iZotope RX 9: From 0 to Hero

1. Introduction to the course: Hi and welcome to the isotope our X9. As you may already know, RX is the leader of the audio restoration and repair software on nowadays market, and its capabilities are mostly restricted only by our imagination. This is one of the must have and go-to applications of every recording and mixing engineer, interview and audio book editor, self-managing YouTuber, influencer, podcaster, voice actor, and so on. During this course, I'll be using the latest release of the program, which is currently nine, and the advanced edition which has the maximum number of modules. I also want to note that isotope RX can work both as the plug-in integrated into your DAW and as a standalone application. In most of the DAWs, we can access only a certain number of modules of the RX. For this tutorial, I'll be working in a standalone application which can showcase the full capability of this product. However, the RX modules you can access from your DAW are the same ones that we will learn here in this series. We're going to take a detailed look at every setting, module and function of this powerful tool. While going through, we will discuss an actually try every component of the RX suite on various problems that audio recordings can possibly have. You also have access to all of the audio files used in this course. After every chapter, you'll be option to pass the quiz about the learned material so you can check yourself. And the last episode of the course is dedicated to applying our knowledge in practice, we will be facing real-life example which contains a variety of problems, discussing and applying different solutions for every issue. Thank you for your attention and let's get started. 2. Interface navigation: When you open the program for the first time, you won't find many similar interface elements as you can find when switching from one DAW to another. And if you think the interface looks confusing, try to drag and drop here your first audio and you'll see what's real confusion looks like. But although it may look unusual, everything here is actually pretty simple, reasonable, and very convenient. I'm sure by the end of the course, you won't find any trouble navigating through this app. So let's start from top to bottom and from the left to right. At the very top we can see a pretty standard panel containing the functions that are mostly doubled in the main interface area. Or it can be called by hotkeys. By the way, this menu is the easiest way to find out a hotkey of a certain function because it's written right in front of it. Of course, we will run through these functions in the next chapters. So for now let's move down. This area is where we're gonna see all the open tabs. Every new audio file we add in the current session will be open in a new tab. Just like this. We can combine our tabs and process them together using this composite view switch. Although the audio files in these tabs must have the same sample rate for them to be combined. Here is also a repair Assistant, which is a pretty smart analyzer. It's looking for problems in our audio and suggesting up to three kinds of solutions that we can choose from. We will get back to it later as well. Here we have a switch that changes the view of the audio from stereo to mono. Note that only the representation is changing. If you have a stereo file and you switch the view to mono, the signal will remain stereo, but we will just see two channels combined in one. So when we need to process two channels simultaneously, we switched to mono view and see a bigger image of our audio. This is the regular timeline when we zoom in and there's no space on the screen to show the full picture are. Once we can just drag the slider from left to right and move the view horizontally. The more we zoom the smaller slider becomes. We can also use these arrows to change the amount of Zoom. This is the representation of our audio itself. By clicking on l and r bars, we can enable or disable a left or right channels. And if we, for example, disabled the right channel, we will be hearing only the left one and it won't change if we switch to the mono view. Once again, this button just switches the view and doesn't affect our audio in any way. Here we can see frequency and loudness bars that they both meant to change the visualization of the audio. I'll talk about it in detail in the next video. So let's skip it for now. And the very right panel is where all the RX modules are located. Every module is meant to fight a different kind of issue. For example, the clip helps to remove distortion from the sinewave picks going beyond 0 decibels. Ds reduces sharp hissing sounds and so on. This is the heart of the RX, its biggest and the most important part. So we will spend a significant part of the course discussing and trying each of them. All the modules are divided into three main categories, which are repair, utility, and measurements. By using these arrows, we can expand or collapse every category. We can shorten this list by choosing the needed a subcategory From this list. Module chain allows us to create a set, a chain of modules that we can apply at once by clicking this render button only saves tons of time when we need to fix one issue in the different parts of the audio. This arrow expands and collapses the modules panels itself. Under the display we can see the timescale. We can change the time format by right-clicking on it and choose between samples, seconds, or frames. Using this slider, we're moving from the waveform to the spectrogram view of our audio file. As I said, we will learn about it in the next video. This whole panel contains tools responsible for zooming and selecting regions of the audio. I prefer to talk about these tools in the episode where I'm going to explain this whole waveform spectrogram representation. After the selection tools, we have the clip gain icon and enables this kind of volume line that we can change in any way. We can precisely raise or lower the volume on any part of the track in the needed amount. This corner area is dedicated to playback and recording tools. Here we access the same menu as we saw when right-clicked on the timeline. These are the play head location numbers. Play head is this playback marker. If click on space here, we automatically place the marker to where we clicked. We can also drag it manually to play forward or backward. By the way, if this button is enabled, play head follows playback. Then when we play something, this play head will move along with the playback and it stops right where it is. If we press Play, the playback will continue from the same point it stopped the last time. But if this function is disabled, the play head doesn't follow the playback. And after we stop the playback, the play head, this yellow marker is coming back to the place where we placed it manually the last time. Like this, I place it here. I play a little stop and it's back to where it started. The input monitoring allows us to hear what we're currently recording. For example, I'm recording my singing and hearing myself in headphones. This is the record button itself. Rewind to the beginning, play stop. Play frequency selection. I'll also explain it in the next videos. Loops selected part and play head follows playback. This is what I explained earlier. Moving gone. This is the loudness information column, left and right channel. The very left edge is infinite silence. And to the right, negative 40, negative ten decibels up to 0. All the pics beyond 0 decibels will be marked red and we will see the amount of decibels that goes beyond 0. Let me rise the volume curve. And I need just to click on it to reset this value. Next is our play head position and length values of the selection and the current region that fits in this window. I zoom-in, zoom-out and it changes. We can also edit all of these values from here, but it's not that comfortable as manually moving the playhead, selecting the part, zoom in or out. The same width, the frequency and cursor location. At the very right corner, the history is displayed of all the actions in the current session. We can choose any event in this list, any action we did and restore our audio to that point. For example, I apply some EQ to some selection. Then I level something. Now, let's say I want to come back to the state where I didn't apply levelling, I simply choose EQ in this list and that's it. Yes, In this case, I could just press Control Z or Command Z and wall one step back. But if I did many actions using this menu, lets me skip many steps and immediately come back to the state I want. 3. Waveform-Spectrogram variations: Now it's time to discuss the RX is audio visualization. By default, after we load the audio file inside the app, we see the combination of the traditional blue waveform that I'm sure most of you are already familiar with. The orange spectrogram. We can adjust the amount of each of them by dragging this slider. By moving the slider to the very left, we can see only the waveform of the audio. Vertically represented the loudness, and horizontally the time from left to right. This is the classical two-dimensional representation of the audio information we can see in our DAWs, media players and so on. We can see that this part is louder than this one. This peak here is shorter than this one, and so on, so forth. What we miss here is any frequency information which is extremely important in audio restoration and repair. That's why isotope introduced a brand new three-dimensional spectrogram view. Let's move the slider to the very right. Now the time is still represented horizontally from left to right, but the vertical dimension is now dedicated to frequency. At the very bottom are the lowest frequencies and accordingly, the highest frequencies are located at the very top. You may already noticed the gradient between the colors. This is the loudness. The brighter the color, the louder this part is. The very black color means there's no audio information at all. Here is our frequency scale. On the very bottom is everything below 100 hertz, and on the very top is everything above 20 kilohertz. Now if we take a closer look at this scale, we can notice that the range between 100 hertz and one kilohertz is much bigger than, for example, between 67 kilohertz. This means that this scale is not linear, but his most practical because it's adapted to the sensitivity of our ears towards different frequency ranges. Accordingly, the range where we hear more details is bigger on this scale, so we have more control over it. However, this is just one of the five scale variations in isotope RX. We can see and switch between them here, right-click on the frequency scale. Let's take a look at the linear scale. This is the most equal frequency view here. As we can see here, is the same interval between the same frequency ranges. Now, the next one calls Mel. This is the default scale we were looking at in the first place. There is not a big difference between Mel and bark. Both of them are based on the humans frequency perception, but bark is a bit more focused on low frequencies. This is how the low frequency register looks like in Mel. This is how it looks like embark. We can see that bark has a little more gradations here. The last two scales, log and extended log, are a lot more low frequency focused. It's easier to trace some hum or another type of low frequency noise with their use. It's also possible to turn this frequency scale into a piano view, which can be helpful for those who work with music. If we scroll while pointing on the scale, we trigger this slider and we zoom through the frequencies. And it's different from what if we just scroll on the main area or use this slider? This is the general zoom through time, and this one is a frequency Zoom. As you can see, they work a bit differently. The next up is the magnitude scale. By dragging up or down, we can adjust the dependence of the decibels from this color scale. For example, let's focus on this negative 20 decibel mark. Right now it's located in front of the bright yellow color. It means that all the information here in the loudness range of negative 20 decibels will be highlighted with bright yellow. And if I drag this scale down, this information becomes darker. Now all of these orange highlights that left are the loudest fragments of this audio. So basically this is our loudest sensitivity. If we need to focus on the loudest parts only we scroll this scale down until the whole background. All our room tone, background noise, ambience fade to black. And the other way around, if we need to focus on the background and other less noticeable sounds, we make everything brighter. But please note that this doesn't change the volume in any way, only the way we see things here. The same thing is with the waveform view. Let's move this slider to the very left. As I already mentioned here, we don't have any frequency information, which is why we can see only the amplitude scale here. Now you see how informative and variable this view is and you can easily set it for your needs and comfort. This three-dimensional representation allows us to get main information about the audio even before we play. Let's take a look. I loaded here a couple of audios and let's imagine I have no idea what are those. I could see these repeatable lines with a specific distance and interval between each other. And if I zoom out, I can see that not only the lines are repeating, but the whole parts of the audio. If I open this piano roll, I'll see that all the lines are right across the piano keys. Needless to say that this is a musical piece and all these lines are notes. These ranges are parts of this composition, probably the verse chorus bridge and so on. These parts have a full range. This part obviously doesn't have a base and drum kick on it. Now let's switch to the second audio. Here. I don't see any repeatable parts. All the information and all the pauses are absolutely random and unpredictable. This is definitely not music and all this dark blue sand, The audio is a pretty loud noise. It starts from the very bottom and goes up to approximately five kilohertz. This is not hum or buzz generated by some device. This can be ocean waves, wind, rain, busy megapolis ambiance and so on. And of course, we will learn how to deal with those in the upcoming chapters. But for now I want you to understand the importance of visualization, which helps us to instantly tell what kind of audio we're dealing with and to detect some of its biggest issues. 4. Zooming and selection methods: This episode won't be very long as it's dedicated to zooming and selecting tools that have a pretty straightforward meaning. All of them are located on this panel that starts from already familiar to us opacity slider. These plus-minus buttons perform the regular zoom-in zoom-out function. For me, it's much easier, just a scroll with the mouse. But if you're gonna use these buttons, just keep in mind that the Zoom will be directed to the selection like this. If there is nothing selected, zooming will be directed to the play head. Next up is the zoom to selection. This one seems more useful as it zooms the selected region until the scream space. Zoom out to show entire file, pretty obvious. It resets any Zoom we have for us to see the whole file from the beginning to the end. Zoom tool is the same. If we would just scroll, we click here, it zooms here, click here, zoom here. Using this hand, we just drag the audio left, right. The other way to move horizontally is to hold Shift and scroll. If we check this instance process box, we can choose one of the five functions here. And this function, this module will be applied to the selected part instantly. For example, I choose here fade, and now every time I select something, it applies fade to my selected region right away. And it will apply and current setting of the module. Right now it fades out, but I can change it in my main fade module to fade in. And now it's instantly applying fade into my every selection. Very useful when we deal with the same issue through the audio. And if we need to listen to the piece before we process it, we just hold Control or Command click and drag it to the right to listen. Like this, I hold Control, click and drag. I hear that I want to add fade in to the beginning of this piece. I just select and it instantly process. The same with the other modules from this list. I'll turn this off. Moving on and we see three similar icons. These are the main selection tools when it comes to monitoring. The first one lets us select the part with the whole frequency range included. In this case, we can adjust the length horizontally, but we can't adjust the frequencies vertically. This helps when we need to listen and process the whole frequency range at the same time. The second icon lets us select not only the period, but also the specific frequency range in it. Now we see this cross instead of the cursor, we simply select the range we need. We move it in any way, changing its size, do whatever we want to listen to the selected frequency range only we use this button. It's enough to click on it once. And the next time we press Space on our keyboard, it would automatically be bound to this button. The third icon lets us select the frequency only through the whole audio. Here we can't select a specific part, but we can move it up and down and change its width. This kind of selection helps to find some static problems that go through the entire audio. Some sort of HM, resonant frequencies, buzz and so on. The next group of selectors is meant not much for listening, but for direct processing or removing selected parts. These tools might be familiar for those of you who use Photoshop. With the use of the Lasso tool, we're drawing the edges of the part we want to select. And let's say I selected a bit too much, hold Option or Alt and draw the part I want to deselect. That's it. So let's just say I want to remove this part. I circle it and press Delete. Done. The next brush tool speaks for itself. Unlike selecting the edges, like in the previous lasso tool, the brush is drawing the body of the selection. And here we also can de-select unwanted parts by holding Option or Alt and drawing it. The size of the brush can be adjusted by holding command or control and scrolling. The Magic Wand automatically detects the region of the selection. It helps when we have a clear region we want to select. For example, this one. In this case, not only it gives me a few seconds, but it also selects this part more precisely than I felt were drawing this election manually. And after we select a piece, we can use this tool. So the RX will try to find up the ten harmonics of the fundamental harmonic to de-select any kind of selection, press Command Plus D or Control plus D. This clip gain tool we already covered while the interface navigation walk-through. 5. Top panel: After we became familiar with the RX interface, it's time to take a look at what's hiding inside these top menu tabs. Some of these functions don't need many explanations, but I feel that this course will not be complete without this walkthrough. Anyway, for me not to make this episode completely boring, I'll try to skip the most obvious functions and the ones that have shortcut icons in the main interface. Let's start from File New. If we want to start one more session in the new tab. New from clipboard, if we want to work on some part of the current audio separately, we can simply select the part we want, copy it, and then choose new from clipboard. It automatically opens the new tab with this selected piece of audio. Open save, save, as these are the obvious common functions I'm sure everyone is perfectly familiar with. Next, save RX document. In this case, we saved the whole session with all the settings and processing, but not the separate audio file. Unlike the regular Save As if we choose Save As we only save the audio file from this session, we can instantly overwrite the original audio file, export the whole audio from this session or only a selected piece. Now, export regions to file. What are the regions in RX we can place here are markers. Just press M and you'll add a marker to where is currently your play head. We can add any amount of markers and it helps us to locate important parts of the audio. We can also mark the whole selection and it would call as a region. I select something, press M, and it automatically adds a region here. If we press Option plus m or Alt plus m, We enter this menu where we see all our markers and regions. We can edit their names, positions, import, export, add, remove, and so on. We can also remove them by right-click or by just holding Option or Alt and clicking on them. If we choose this export regions to File option, we have our regions and the ranges between the markers export it in different files. Here we can close our current tab or all the tabs together. Export screenshot and the history, which is a comfortable way to showcase your workflow with your colleagues or clients. And we can also restore one of the recent files from this list. Next up is the edit tab. Here we see a bunch of common functions that I believe don't require any explanations. And here is Paste Special. These are the options of how do we want to paste a piece of audio from buffer to our existing audio file. For example, I select any part here, copy it with control plus C because I'm on the PC, but for Mac users it's Command plus C. Now I want to paste it also here, but in a different place. Let's say here, I can choose how you want to paste it by default. If insert, this piece will be placed right between the left and right sides of the cursor. Like this, you can see the audio became longer in total. If I choose replace, my piece will replace the audio from the right side of the cursor in the amount of its length. If mix, then accordingly this piece will be mixed with the main audio. If invert and mix than the audio will be inverted in the clipboard and then mixed with audio in the project. This is useful when you want to compute differences between two signals. To Selection pastes audio from the clipboard only within the selected bounds regardless of the copied audios length. And the last option pastes only the clip gain information from the copied piece. Next, remove the selection or repeat selection. Select all these two options, revert the selection. This is when we select some part. After reverse, everything else is selected, accept this bar. The difference between these two is that invert selection frequencies don't invert anything else but the frequencies of the current selection, unlike invert selection when everything is affected, select harmonics is already familiar to us that we were talking about in the previous video. Here it is. Now these two functions are very useful. They let us set the beginning and the ending of the selection while playback. It's much more comfortable to use hotkeys for these, which are these square brackets. So for example, I'm listening to the audio and I noticed that some noise occurred. I instantly press the opening bracket. When the noise disappeared, I pressed the closing bracket, and this range is selected right away. It's very convenient. Delete selected part. Trim to selection will delete everything except the selected part. Glick and the selected region is the only one that's left. Here. We can enable, disable Snap and choose what we want our selection to be snapped too. We can find the part that is similar to what we have selected. We click Find all. We see that more regions are selected. Let's listen. This is our original selection. This is what RX found. The same part. If we have a lot of similar events found, we can switch between Previous and Next once we can the level on how similar should be these parts that were searching for? Right now, it looks for everything that is at least 50% similar to what we're searching for. When RX finds it, we can click here and instantly get markers on all the similar parts we found. This is how we add a new marker or the region. And these are all the zooming and selection methods we discussed in the previous video. All the RX preferences we will cover in the next video. Moving on to the View tab, which is pretty short. Collapsed module panel is the arrow here. Time format is what we're setting here. Second samples frames. Here. It's also a shortcut to the frame rate settings, but this is also a part of the next episode. We can enable or disable, follow play head function and choose its mode. When enable and page mode, this piece of audio, we'll switch it to the next one only when the playhead reaches the end of the current screen zone. When it's enabled in continuous mode, the screen is moving together with the play head. The play head is always in the middle while playback. Effect overlays, these are the options of two modules, and it's better to come back to these options while learning the modules themselves. So you will better understand what are these parameters for. Clip gain is the volume curve that is also activated by this icon I have shown before. Show channels separately is the mono stereo view switch that changes the view, but it doesn't change the playback. And the last position here, spectrogram settings. I also included it in the next video. We will cover all the settings and preferences at once. The modules tab contains all the elements from this panel, and of course we will discuss them one by one in future videos. The transport tab contains all the playback and recording elements from this panel that we discussed before. Up next is the window tab. Here is the module chain that we can also find here. The module chain is where we can add a few modules and process the audio with all these modules with one click on me. But I'll be showing it later when dealing with modules already. And then goes a batch processor, which is in my opinion, one of the coolest things about RX. It can also be opened by Command plus B or control plus b. What's that? This is a sort of container where we drop multiple audio files and process them at once with our module chain. Why do we need this if we can just drop our multiple files directly into the main RX area. Yes, we can drop here a few files, then switch it to the composite view where all these files will be represented as a whole. And then apply one module or the module chain to process all the files at once. But first of all, here we can load only up to 32 files at the same time, which means there will be 32 open tabs. Second of all, to switch them into composite view, all of these files must have the same sample, right? And third of all, we won't have flexible export settings in this case anyway, this is where the Batch Processor helps a lot. In the book processor, we can load any amount of audio files with different settings, sample rates, and so on. We can drag and drop audio files directly in this window or click here. Then we choose what we want to do with all these files. All of the RX modules can be found in this list. And of course we can select one of the presets. The suggested module chain will be automatically loaded. Let's say I want to clear dialogue. Here it is, and here are the modules for this purpose. By default, I can change and set them in any way I want. Let's say I'm happy with this module chain. Next, what I need to do is to add an export option. I choose a location where these files will be exported. Then I choose a file format and its settings. Here I can add a prefix, a text it before the file name. Let It Be test. Now we can also add one more export option. The first one was wave, and let's say I want all these files to be also export it into MP3. I choose here MP3 choose the best quality. I'll leave the same location and I can also add the suffix. This is the text that will be after the file name. Let it be clean. I click process. And now these files will be processed with the modules and exported with these settings. It may take awhile, depends on how many files you have here, how big and heavier processing chain here, how many exporting options you set and how powerful your computer is. Now it's done. And when I opened the destination folder, I can see all these files and wav and mp3. Here is their test prefix and clean suffix. This is a very powerful and unique tool that saves an incredible amount of time when working with multiple files. The next three elements here are part of the module section. We will check them later in this course. Although we've already discussed this markers and regions menu earlier in this video, reopen closed windows is also very helpful function, especially in busy sections when we're dealing with multiple modules open at the same time. This is what usually happens to me when I'm working on difficult recordings that have a lot of issues. My layout looks like this. Here I have mouth D Click. Here is deconstruct, then voice, the noise, gain, fade, and so on. Obviously there is not much space left on the screen, especially when working on a laptop. So pressing Control plus Alt plus W on PC, or Command plus Option plus W on Mac closes and reopens all these windows simultaneously, which makes this function one more time saving hero of the RX. In File info, we can see any details about the audio in the current tab. Here is the list of all tabs there are currently open. We can switch between them here by clicking on their names or using these Previous and Next elements, which is definitely much faster to do with the use of hotkeys. It will just be the same as if it were just clicking on these tabs directly. And finally, the last Help tab where we can access the different text and video manuals and tutorials. Opened the list of keyboard shortcuts and check details about your current RX instance. 6. Settings: Before we proceed to the biggest and the most exciting part, the part I'm sure you're watching this course for, I'd like to dedicate this episode to go through our EC settings. This is gonna take a while and all this editing elements have a very clear drop-down description. So if you feel this description is enough informative for you and you want to start learning modules. Feel free to skip this episode. Besides, you can come back to this video anytime you feel the need to. We can find the general RX preferences at the very bottom of the edit tab or by clicking command plus comma or Control plus comma. The first tab calls audio. Here we choose the type of our audio driver. And if we mark this checkbox, this audio driver will be used by RX only while playback. This is useful when our audio driver isn't multitasking. And for example, we often switch between RX and DAW, which also uses the same driver. This is what device we will use to send the signal inside RX for recording. And this is what device will be used to send a signal back to our speakers or headphones. Preferred layout, this concerns only multi-channel files. Here I have loaded 5.1 forest recording, which consists of six channels. When film is chosen, these channels are sorted in the following order. Left, center, right, surround, left, surround right, and the low frequency effect, which is also known as subwoofer. But when we choose the SMPTE standard, which is the Society of Motion Picture and Television Engineers. Then we have these channels in a slightly different order. Left, right, center sub surround, left, surround right. The number of buffers and buffer size are responsible for the latency. If you're experiencing uncomfortable delays, Rog recording your playback, try to lower these values, but get ready for the increased CPU usage. Stereo down mix is the algorithm used to playback multi-channel files on stereo and mono devices. And here we can choose one of those. For example, to mute low-frequency effect and play without it, or increase volume on surround channels on three decibels and so on. All these options here are pretty obvious. The test tone is a frequency generator. Also very useful thing. We can choose here a preset with a specific frequency or noise. If we choose the custom frequency, we get access to the slider. This pure sound wave helps us to train our ears and things are becoming a lot easier when we can recognize frequency ranges on the ear. This amount of decibel will be reduced from every audio that we collapse in a composite view. This helps us not to shock ourselves when we decide to press play in composite view and all of the combined audios will start playing together. This output gain fader is just a master volume. This is how loud we want our playback to be. Next up is display. Here we can turn off tooltips. Tooltips are the description lines that pop up every time we point on some function. Display cursor coordinates in status bar, it means this panel. If we turn this function off, we won't see any cursor data here. Show analog waveform. If we enable it, we see that the edges of the waveform slightly changed contrast. If the analog waveform is wider and some parts we will see some red edges. This is the way RX is trying to predict the behavior of this sound wave when what is converted from digital to analog, which is meant to make us aware of possible clipping, offload waveform calculations. It's enabled by default. And when we load some audio are x automatically calculates the waveform, which is slower the loading process. But if we turn this option off, we can load the file faster and already use it while the calculation is processing in the background. Waveform interpolation order. Roughly speaking, this is how smooth and precise we want our waveform to be drawn. The difference can be seen only in deep zoom. If I select the lowest value. All the individual samples will turn into these squares. If high value. Like this. The brightness of the interface and the opacity of the RX windows. All the windows except the preferences Windows. Let's reset all the changes here and move to the next tab called keyboard. It's only about hotkeys. For example, I want to apply ds by pressing Shift plus D. I searched for dS related functions here in the list or using this search bar. Here is my apply dS at the very beginning of the list. Here we see that this column is empty, which means no hotkeys were assigned to this function yet, I click here and press my Shift plus D on the keyboard and click aside. That's it. Now every time I wanted to apply ds or to any part of the audio, I just press Shift plus D and it'll be processed with DSR. I can easily remove my assignment and leave this function on assigned or choose a different key combination. Let's say I want to apply ds by pressing Control plus S. I click here again and press Control plus S. But now I see that this combination is already in the US. It's bound to the save function. And if I press a sign, this function will be replaced and control plus S will be assigned to the DSR. If I don't want to rewrite this hotkey, I click here again and choose the different combination that isn't assigned yet. Here we create import or export our hotkey presets. Reset everything again. Next up is the miscellaneous tab, miscellaneous settings. This path is where our project data is located. By clicking on this arrow, we can change, explore, or reset to the default path. If we press change, we need to choose where to locate this folder. If Explorer, we opened this project folder and this is where all the cash, although temporary data from our current and previous sessions are located after we work on heavy sessions log files, a lot of processing, we may end up with a few gigabytes of the temporary data in this folder. So we can delete this session folder if we don't need to come back to it anymore. Timescale frame rate is how many frames per second we will see on this scale when our time format is set to the time code or the source time code. Here I have chosen a time code and here are my frames. This value is how many of these frames I'll have per 1 second. These two parameters are connected to the Special Paste modes I was explaining in the previous video. And here we choose what paced mode will be used by default. In other words, how the copied audio from the buffer will be pasted in the general work area, will it just be inserted in the current audio file? Will it replace the needed range? Will it mix with it, and so on. So the first function is where we choose the default paste options for the full bandwidth selections. It means here I choose held the piece of audio will be pasted by default when I use the time selection tool for exemple, I want every time I copy paste something for it by default, replace their current Bart. I choose here, replace. Now I use the time selection tool to select any part. I select, copy this piece and paste it somewhere here. Also paste here, here, and so on. And every time it replaced the existing piece. If I don't want it to be replaced, I choose another mode here. And this second function is totally the same, but only when it comes to the time and frequency selection method, lasso, brush and magic wand. In this case, I choose only the specific frequency range I want. I copy, paste somewhere here. And it also replaces the existing piece. Because here in the settings I have replace. So again, this is the default paste method for the time selection tool only. This one is for all the time frequency selection tools. These two options speak for themselves. We leave them enabled if we want our audio and all the windows from the previous session to be restored every time we launch RX, automatically open files ending with dot L and dot R S split stereo. This option is very interesting and it enables by default, let me demonstrate how it works. So I have two mono audio is here. If I drop them into the RX interface, they will be open into different tabs. Nothing special. But if I add at the end their names dot L and then give them the same name. Let's say this one is guitar dot L, and this one is guitar dot r. In this case, these two mono files will be automatically recognized by RX as two parts of one stereo file. And accordingly they will be placed in one tab as the left and right channels. Recall Selections during undo redo. This is when I, for example, select some region, process it, remove the selection. Now let's say I play it again and I decide to undo the processing and apply something else by default. After I undo, the selection remains right there where I was applying my previous profit setting. And now I can easily use another processing without the need to re-select. It works both with undo and redo. Play only selected channels. Here it is. If I click here, only the left channel will be selected and only the left channel will be playing the same with the right one. And if I disabled dysfunction, both channels will be playing whether they are selected or not. Next, if we leave this checkbox are RMS root mean square value will be calculated using the International AES 1 seventh standard. If uncheck RX will be using its internal algorithm. Pre-roll and post role during preview. This amount of milliseconds will be also played during the preview or Barstow elected part. We didn't learn any modules yet, but I'll quickly say that most of them have a preview option. This allows us to hear the processing before we render it. So by default we have 1 second here and here. This means when we select some peace, open the module, we want to use tweak something and click Preview. It will start playing 1 second before the selection and stop 1 second after the selection ends. But these 1 second here and 1 second here will be played unprocessed. It helps us to instantly here the difference between unprocessed and processed sound. And here we adjust the amount of time we want to play beyond the selection. Or if we want to preview the selected range only, we just have to drop these values down to 0. And the last in this tab, selection feathering is how smooth we want our processing to start and end. Let me drag this slider down to 0 and let's take a look. I will select any part, and for this example, I'll use the Gain module. It's a simple module meant to increase or decrease the volume of the selected part of the audio. So let's say I want to reduce the loudness of this part and negative 90 decibels, I drag this slider until negative 90 and click Render. And now we see that this loudness processing affected the audio abruptly within the selection. If we want these edges to be less sharp and more smooth, we're coming back to this function and changing value. Let's go to the maximum which is 1 second, and let's try now. The same selection, the same game module with a negative 90 reduction setting. Click Render. And we see that the edges become much smoother. This is how it will be with any type of processing. It's just I used this game module for the obvious visual result. The next tab is dedicated to authorization and updates. Here we can authorize and Unauthorized the RX instance if for example, we're using it on a temporary device and then choose what type of demo will be left after we log out. This link will lead us to the isotope product portal download page. Product portal is an application which is a sort of control center where we can manage all the isotope products currently installed on our computer. Here is how it looked like. There are three main tabs where you can find all of your installed isotope products and all the products you can try for the ones you've already tried but didn't buy yet. You can install and upgrade them right from here. Also, here's a bunch of useful links leading to the isotope resources. The next tab called Plugins. And this is the one of the most powerful things about the RX. Yes, in isotope RX, we can work with our VS T2 or audio unit plug-ins without even the need to open our DAW. Here we can see all the plugins correctly scanned by RX and available for use. We can enable disable them. And here is where we can add or remove the path where RX will search for the plugins. And if we add one more path, we need to restart the application or simply press rescan for the plugins from the new path to become available in RX, this checkbox will automatically sort all the plugins that have at least one same word in their names and not here, but the plugins module manual, where we will automatically run our plugins, of course will cover it later. And finally, the last tab where we can mark the modules we want not to be suggested when running the Repair Assistant. I'll run the Repair Assistant to ask for some suggestions about the Quality Improvement of my audio. Rx scans the audio and if something here has checked, the assistant will skip the module when suggesting me the solution. Now that we're done talking about the general preferences, it's time to move to the spectrogram settings. We can find it at the very bottom of the view tab or by right-clicking on the display. Here's the regular preset tab where we manage our presets here, or we choose one of the four algorithms that will be used to calculate and represent our waveform. The first one, regular STFT is the fastest but less precise calculation algorithm. When I'm saying fastest, I mean how fast our waveform will be drawn while zooming. This suits the cases when we can sacrifice surgical precision for the fast workflow. The second algorithm, auto adjustable STFT is a bit smarter than the previous one. Here are detailing depends on Zoom type. If we zoom in horizontally, we will see more detailed timing information. And if we zoom vertically, then accordingly it will be focused on the frequencies. Multi-resolution is based on humans Loudness, Frequency perception. This algorithm shows us clearer frequency details on the low end and clearer timing details on the high-end. And the last algorithm here is adaptively sparse, which provides us the most detailed picture of both in the time and frequency zooms, but require more time to draw the spectrogram when zooming. Enabled reassignment, this checkbox enables a special algorithm that sharpens tonal components of the audio. This is how it looks like without reassignment. This is how with reassignment. But this function provides significant results when it pairs with frequency and time overlaps. These are two oversampling values that lead the reassignment to the time or frequency direction. Simply saying when we work with this music or any tonally oriented audio, sometimes it helpful to enable the reassignment and to try two of these values. Frequency overlap will provide us with more pitched details and time overlap with more information about transients. This also takes some time for these functions to calculate the war values we have, the longer it will take to compute FFT size. What's FFT? Fast Fourier transform, a procedure for the calculation of a signal frequency spectrum. The greater the FFT size, the greater the frequency resolution. Ie notes and tonal events will be clear and larger sizes. But the higher the FFT value, the less sharpen our time information is. Let's take a look. I'll increase FFT to the biggest value. And when I zoom in, I'll have more control over the frequencies. In less control over the time. Hold Option or Alt, and click to reset any value to default window. This is where we choose the mode that will be eliminating the signal leakage. What does it do for us? Slightly different frequency representation on different zoom levels. I usually work on the default window mode, which is hand. And in case if I hear the problem but can't see it, I'll try the other modes. The color map will free to choose any color combination we want. And here are no step stones like losing performance or choosing between time and frequency quality, Nothing like that. This is only for personal convenience, though some of the combinations here might be too tiring for the eyes during long sessions and some of the combinations can be more informative. In some cases. Here we can turn off this color slider. And if we turn this function off, we will get a bit faster render, but will also lose a little quality. So I wouldn't recommend you turn it off. These two sliders are connected to this color scale. It's the same if we will be tweaking the scale directly to the main area as I've shown before. This slider allows us to specify the length of the visible spectrogram that will be calculated with full accuracy. It's 90 seconds by default. So when I zoom out and look at my audio, the spectrogram image is not as accurate as when I zoom in and my visible part becomes 90 seconds short. At this zoom level, I can only see 90 seconds of the audio at a time. So the spectrogram is already drawn with full accuracy. Here we can adjust the amount of the cache. This is the temporary data that RX will store on our computer in order not to recalculate every time information that wasn't changed. It helps to speed things up. So I recommend not to be too greedy here. 7. Ambience Match: Now that we're done with all the interface details, audio visualization variations with all the menus and settings. It's finally time to start learning our x modules. As we can see, all of the modules are divided into three main categories. Repair contains modules that are mainly intended for cleaning and repairing audio. Utility includes modules designed for fast and flexible mixing and balancing of sound. And the measurement section is where we find tools for statistics and navigation. We will start from the repair section and the very first module in it calls ambience match. Let's open it up and quickly rock through the general interface of the RX modules. These are a few functions that we will see an almost every module in RX. This is the regular preset window here we can find the predefined templates and choose what suits us the most. When we select some parts and press learn the module will analyze this current piece and search there for the problem or solution. After the module learned the selection, it sets parameters that fits the best for the learning part. Then goes the Preview button that we've already slightly discussed. It lets us here the processing in real-time before we apply it. Then the Bypass option. This works great when we preview the processing, allowing us instantly switch between the processed and unprocessed sound. The compare window. We can add comparison Options and switch between them to listen and see the changes on display. And render is when we want to apply the current settings to the selection or if nothing is selected, then the whole audio. Okay, so let's focus on the ambiance match module. What's it for? What is the ambiance in general? This is the type of environmental sound that surrounds us everywhere in our house and the traffic, grocery store, Beach office park everywhere. The ambiance always presets in every recording, even made in well acoustic treated studios. But in this case, it'll be too quiet for us to hear or at least to bother about it. Let's go to the example. The person decided to record one of his weekly podcast episodes at the seashore. He completed the recording and when he came back home, he realized that he forgot to include something important in the his episode for any of the reasons he can't or doesn't want to come back to the same beach to record only a few sentences he missed. And he decides to record it right in his bedroom and add it somewhere in the middle of its main audio. Let's listen to the short piece of how it sounds like. So being in a place where people will just come up to me and say, hey, what's up? Hey, can we have a shot? It's ten o'clock in the morning, but you can drink beer, right? Yes. Yes. Like I always used to drink this until I was about 2122. This was my drink of choice. I still drink it now sometimes, but a lot of the time it gets too sweet for me. I admire. They have lots of people who will just come up to me and say, Hey, I'm actually quite shy. As we hear, the bedroom part has much better quality. It's very clear and there is much less noise. But we can also hear the part that sounds pretty unexpected and even a bit weird when it starts about the seashore recording. What can we do about it in cases like this, there are always two ways to make a different recording sound alike. The first option is to try to remove the ambiance from the noisy audio. And the second option is to add ambiance to the clear audio. In this example, we will go for the second option. Why? Because in this case, the major part of the audio was recorded in a windy open space and the bedroom pieces much shorter. So since our task here is not to make the audio as clean as possible, but to eliminate the difference between the two recordings, it will be much faster and easier to add an ambiance to the clean part besides the wind here is way too strong. So if we try to clean it up, this C-sharp part won't sound as good as the bedroom part anyway, the artifacts will occur and the overall quality will be lessened. So how can we add the same ambiance to the clean recording? If only we had a part of the recording without voice, we could simply copy, paste and mix it with the clean recording. But in this example, all the ambiance also includes the voice. This is when the ambiance match comes to help. First, we need to specify the type of ambiance we're dealing with. The static mode here is the perfect for some permanent background noises such as forest, room tone, rain, and so on. The complex mode is more suitable for the ambiance where something is always changing. Busy street, for example. In our case, we're dealing with the static ambient. So accordingly we're choosing the static type here and the next or selecting where this module were taken example of the ambiance. Note that a static mode will get an example of background noise only. It will not learn from any dynamic elements of the audio. So now we don't worry if there is a voice in our selection, only the background will be learned. I will choose a large enough part to get more diversity. I click Learn, and here we see the ambiance our X generated for us. We can compare how it looks here. All these bright elements or the voice. And here everything is pretty permanent, smooth and static. If we take this checkbox output ambience only, we can here only this piece of the ambiance we're going to use. As we can cure. There is no voice only pure background noise. This is just what we need. The last left thing to mix is the seashore ambience with the bedroom dialogue. We can use this slider to change the volume of this ambiance we generated, but let it be by default. I select the part with the bedroom dialogue and press Render. That's it. Let's listen. Being in a place where people will just come up to me and say, hey, what's up? Hey, can we have a shot? It's ten o'clock in the morning, but you can drink beer, right? Yes. Yes. Like had I always used to drink this until I was about 2122. This was my drink of choice. I still drink it now sometimes, but a lot of the time it gets to speak for me. I admire. They have lots of people who will just come up to me and say, Hey, I'm actually quite shy. Of course, we can still hear a slight difference between these two parts, because even though we layered the same ambience, the voice itself was recorded in a different place, but it's much better than it was before. Besides if we weren't focused on this difference, we probably wouldn't even notice it. Now let's talk about the complex mode. The complex mode is meant for more dynamic, more active and variable ambience types. Let's switch to the second example. This is the recording of a walking toward of downtown New York. We cannot hear many static elements here. Everything is changing. Some random sounds at left, right. We can see that in complex mode, the ambiance threshold slider becomes active. This is where we adjust what will be considered an ambience. For example, if someone's voice is present and it's quite loud and we want to extract the ambience without it, we can try to adjust the threshold. 0 means all the sounds up to 0 decibels will be captured. If negative 20, nothing above it will be captured and so on. Let's try to find voices in this audio recording. This part will do. There was someone who will talk a little. If we learn it with a 0 decibel threshold, the whole original part will be learned. Nothing will be reduced or removed. Now we click Learn with a 0 decibel threshold. And when we click Preview here that the voices are still there. Now let's move this slider to, let's say negative 20 decibels. As we can here. There are no upfront voices anymore, but it also started to sound a bit less natural. In this case, it's sort of a side effect of using a tight threshold. Next, in the complex mode, we get access to these two sliders. The movements lighter gives us control over the most dynamic elements of the ambience. The higher the value, the more original sound is, the lesser the values, the more of these dynamic elements are reduced and replaced with the more static parts of the selection. We can see that some of these elements here are the preview spectrogram. They're becoming gray as we move the slider. When we're increasing randomness, we're letting our x chopped the selection into many tiny pieces and randomize them. We have to be careful with this lighter as it may sharpen the audio, make it sound abrupt, add distorted. Now we can do with this piece of complex ambience the same we did with the static. Choose where we want to paste it and press Render. I always used to drink this until I was about 2122. This was my drink of choice. I still drink it now sometimes, but a lot of the time it gets to sweep. Ambience match also includes a variety of noise presets. They are playing. John goal, rain, and so forth. These are the common good quality ambiances we can easily use in our projects. 8. Breath Control: Breath control is designed to track breathing quieter it or remove it entirely dependent on our settings. Before we start, I want to note that breath is not always an unwanted noise and should not always be completely removed. There are a lot of situations where breath brings an artistic color to the dialogues or the vocals, making them more alive and emotional. But of course, sometimes breath can be really annoying, too loud, sharp, or repeatable. And dealing with it manually is not what we can call a faster interesting thing to do. Let's go to the breath control module. We can see that the top and the bottom panels here are totally the same as we saw in anti-B and match. So there is no point to repeat, but of course, the main area of the module is different. And let's figure out what we can set here and health of breath control works in general, the module has two modes, gain and target. These are the modes of how breath control will react on detected breathings. In gain mode, the loudness of every found breath will be decreased on its value. It's set to negative 30 decibels by default. So right now, breath control will automatically decrease negative 30 decibels from any detected breath regardless of its original loudness, this heavy loud breath will become negative 30 decibels quieter. This light quiet breath will also become negative 30 decibels quieter, and so on. This mode suits well when we're dealing with breaths that have a constant loudness, especially if it's one person's monologue. The target works a bit differently. When we switch to target, this slider becomes a threshold and it's also set to negative 30 by default. But now it's decibels relative to full-scale, which means the loudness measurement as relative not to the level of current audio, but relative to full-scale. Now everything under negative 30 decibels will not be affected by breath control and anything louder than negative 30 decibels will be suppressed. This indicator in the middle shows us a level of reduction while the breath processing, the use of this slider, we tell breath control how accurate it must search for the breaths with the lowest values, it will react on the loudest and most obvious breaths Omi, and with the highest it will look for any breath possible. Let's get to the practice for this episode. I prepared three audio files. Male speech with a strong breath, female speech with a light breath, and female vocal with moderate breath. And we will try to figure out what settings fit the most to each of these audios. Let's listen to the first audio example. The modeling industry has changed. It's not all about height, size, or beauty anymore. Water are so many things you can do to you guys asking me stupid questions about a guy who doesn't care about you? Yes, it was Betty's in. Here are three quite heavy, obvious breaths which are pretty equal length and loudness. Let's try the default settings of the game mode. The modeling industry has changed. It's not all about height, size, or beauty anymore. There were so many things you can do. You guys asking me stupid questions about a guy who doesn't care about you. Guess Who's Betty thin. It reduces breast volume a lot, but we can still hear them a little. Let's go for, let's say negative 42 decibels. The modeling industry has changed. It's not all about height, size, or beauty anymore. There were so many things you can do. You guys asking me stupid questions about a guy who doesn't care about you. Guess Who's Betty Xin? Perfect. It cut all three breaths completely without any unwanted side effects. Let's try hold a target mode. We'll handle this. Modeling industry has changed. It's not all about height, size, or beauty anymore. There were so many things you can do. You guys asking me stupid questions about a guy who doesn't care about you. Guess Who's Betty? The'm the modeling industry has changed. It's not all about height, size, or beauty anymore. There were so many things you can do. You guys asking me stupid questions about a guy who doesn't care about you. Guess Who's Betty Xin? As we see to achieve spotless results, we have to pull the target level slider all the way down, but it's still provides an excellent result. Now let's switch to the second example, which is more difficult because the breaths here have a slightly different levels, different durations, different distances before and after the words. Let's listen. Oh Honey, you can't possibly be accepted into the contest. Don't you see how short you are? If you're practically a dwarf? Hello. I'm going to be telling, I'm going to be telling you about my crazy rich life. Despite the fact that I was still acute tiny baby, the breaths here are also much softer and quieter than in the first example. Let's again start with the default settings of the gain mode. Oh Honey, you can't possibly be accepted into the contest. Don't you see how short you are? You're practically a dwarf. Hello. I'm going to be telling you about my crazy rich life. Despite the fact that I was still acute tiny baby. We can hear that only the first breath was removed completely without any problems. All the other was whether only a little suppressed or disappear together with nearby letters. Besides, it also removed the letter F from the word life and it cut in half the letter F from the word dwarf. They were mistakenly recognized as a breath. This result doesn't satisfy me, so I'll try to play with the gain in sensitivity sliders. Let's try the target mode now. Oh honey, you can't possibly be accepted into the contest. Don't you see how short you are? You're practically a dwarf. Hello. I'm going to be telling you about Mike, crazy rich life. Despite the fact that I was still a cute, tiny baby. Honey, You can't possibly be accepted into the contest. Don't you see how short you are? You're practically a door. Hello. I'm going to be telling you about Mike, crazy rich life, despite the fact that I was still a cute tiny baby. So what happens here? In this case, it's better to keep sensitivity on the maximum because I hear it started catching more details. But even if I pull the target level slider to the very bottom, it still doesn't recognize those are the last two breaths here. And here. Since there's nothing else, I can tweak the change in the outcome. I'll move on to the next example. It comes in. Here, we don't need to remove these breaths completely, maybe only quiet them slightly. I'd even say this breath level is not critically loud. And there are tones of songs that have a pretty low breath. But we'll try to quiet them just for practice. Negative 30 decibels is way too much for our purpose here. So I'll try to listen to it with negative ten decibels and negative five. If you feel like, you know, just do the sun comes in. When you when you said add speed or you feel like, you know, Target mode. When you add speed or you feel like you got to go. The sun comes in the morning. Yeah. Okay. So among these three examples, I'm completely happy only with the first one. It was absolutely spotless. In the second example, RX manage to handle only like 25% of the problem without any kinds of side effects. And in the case of this particular vocal, I managed to suppress a breath in the price of the harsh endings and slightly chopped beginnings of the breaths. Needless to say that even though isotope provides us with a top-notch software with very smart and flexible algorithms. They're still not always understanding us in the right way to think that we should be aware of when applying breath control with type settings to the tender light breath is that together with breaths, it may also affect sibilant consonants of the speech, especially in gentle Eric kinds of pronounciation. It also sharpens the beginnings and endings of the phrases. It might be forgivable. One our podcasts that has to be roughly edited for the quickest turnaround, but it's absolutely unacceptable when it comes to high-quality vocal productions. For example, sober for needing surgical breath editing and wide control over its feedings and fade outs. Nothing yet can provide as much control over the situation as the manual editing of every breath draws. And RX has all the needed tools for it. We will be covering them in future episodes. 9. Center Extract: The center extract is a module with a quite simple function. It lets us isolate only the mono apart from the stereo signal or the other way around to extract the sides only removing everything from the center. This is the way how we are getting the karaoke instruments before. Since in 99% of all those songs, the lead vocal is located in the middle. We could just cut out the mono and keep the sides only. Of course, it was affecting the kick, bass, snare, and other centered place instruments, which is why we have a music rebalance module which has a much smarter algorithm for these kinds of operations. However, karaoke is not the only purpose of this tool. Let's listen to the example. The problem occurs when decluttering takes over and begins to distract us from what we really want. The voice here is right in the middle and a lot of noise on the left and right sides. To extract the voice here we need to use the keep Center option. And here we see three sliders. When we move the reduction strength slider, we adjust how hard the sides of the stereo signal will be suppressed. The more we move it to the right, the fewer sides we will hear. Artifacts smoothing helps us to find a balance between the noise reduction quality and the quality of the isolated signal. In this example, if I move this slider to the very left, RX will provide me with the raw, well isolated voice, which includes all the artifacts. If I move this slider to the very right, RX will try to remove these artifacts to make the voice more natural, though the reduction quality may lessen. The dry mixed slider regulates how much process sound do you want to get? When it's 0? It means there will be no dry signal. Dry means unprocessed original soul. When it's 0, the entire signal will be processed with its plugin. Let's listen and just move these sliders while previewing. Everything will be easier to understand. I'll press bypass first so we can hear the original audio once more. The problem occurs when decluttering takes over and begins to distract us from what we really want. Now, I'll be increasing reduction strength. The problem occurs when decluttering takes over and begins to distract us from what we really want. The problem occurs when decoder it takes over and it begins to distract us from what we really want. The noises became much quieter, but also the voice lost a lot of its strength and became volume unbalanced. It's very obvious at the end of the phrase when the narrator says From what we really want, why exactly there? I assume it's because the narrator is artistically lowering its voice to highlight the end of the phrase. And the lower frequencies are the more omnidirectional they become. So I assumed since there are more low frequencies in the part of the phrase, the more of them spread around and the less of them left for the center that we extracted. Now let's hear what differences, artifacts, smoothing does. The problem occurs when decoder it takes over begins to distract us from what we really want. When it's all the way left, the voice becomes a little stronger and audible, but loses a bit of quality. And if I move it all the way to the right, the problem occurs when decoder it takes over and begins to distract us from what we really want. The voice becomes a little more natural, but also weaker and less balanced. In the dry mix, we adjust the amount of the original signal to the process one, Let's now switch to keep sides mode. It has two available algorithms, true phase and pseudo pan. The true phase, our X removes the center of the stereo signal and keeps sides and touched. In pseudo pan RX is cutting out left and right channels and then artificial link gluing them together into stereo. Let's listen to how this works. We'll start from the true phase. It works pretty well, but we can still hear some of the voice, although we can't recognize now what he is saying. Pseudo pattern. Here, we don't hear any voice anymore. It cleared completely, filled the quality of the sides is worse. But if I now reduce the production strength, we're getting much better side quality and still don't hear any voice. Awesome result. 10. De-bleed: What do we need the bleed for? What is a bleed in general? In the audio production world, the term bleed or bleeding means leaking of the signal, sneaking one sound into the another. Sometimes this is exactly what we need and we use crosstalk plug-ins to blend signals and so on. But sometimes what we get is unwanted sounds we wish we could get rid of. Let's listen to the example. You could see. It was stronger. We can clearly hear those clicks, those straight snaps through the vocal. And even if these snaps are an actual part of the arrangement, we wouldn't want to have anything else in our vocal track anyway, simply because if we later process our vocal with the reverb or delay, these snaps blended in the vocal track will be processed as well. They will be liberated and delayed, creating nothing but a mess. Let's open the GI bleed module. Now before we start doing something here, I want to note that the bleed requires an example of the noise we want to remove. If we don't have it, we won't be able to do anything here. So we need to provide to Rx2 audio files, the one we want to have clean and the one that contains only the unwanted sound. We want to get rid of. Both of these files have to be perfectly synchronized with each other. For this episode, I prepared two audio files containing v snaps. The first one is 100% synchronized and the second one is totally out of sync and timing. We will see how to bleed works with both of them right after we understand how the module works in general, after loading the main fall we want to clean and the fall with unwanted sound. We should see two open tabs and our X. Here is my main audio and here are the snaps only. In this list, we need to select the file containing only the snaps or whatever you want to get rid of. And before we press learn, we need to make sure we're currently on the tab with the audio. We want to clean this one. Learn. And now we see the spectrogram of the snaps and the spectrogram of the audio we're cleaning. These are the two sliders we're already familiar with, reduction strength and artifacts smoothing. They are typical for our x, so we still need to see a lot of them during the course. Let's keep the default for now and listen to how it sounds after we learned the original bleeding track. Did you could see what I see. Stronger. The result. Did you could see what I see. It was so much stronger than they are no more snaps, but some sort of ducking occurred that the volume is falling all the places we're used to be snaps, let's decrease the strength to very little. Did you could see what I see. It was a stronger, they'll let you know it's not enough. Let's try 0.2. Did you could see what I see. It was stronger. Yes, it works and I can call it perfect balance since the snaps are almost undetectable and the ducking is almost missing. Now let's try artifacts smoothing. Very left. You could see what I see. It was stronger. They're legit thing. And very right. Did you could see what it was so much stronger. Honestly, I can hear only the tiny difference. When there is 0 of the artifacts smoothing, the voice snaps, disappearing a little more, though the voice also becomes a little more sharpened and everything is opposite. When there is a maximum of artifacts smoothing, the voice becomes just a tiny bit softer, but the snaps appear a little more. I'm gonna leave it somewhere in the middle. This is pretty much it. You can do this not only with clicks or snaps, but with absolutely everything. Do you have an example of? But of course, the quality of the result may vary with different types of bleeding. And as I said, the track with one unwanted sound should be perfectly synchronized with the original file. What does that mean? For example, in my audio, the first click starts somewhere here from 0.85 to 0.86 of the second. This is the first click, and we can see it in the click track. It starts perfectly at the same point from 0.85 to 0.86 and the same with every decks click next snap. This example was taken from the Song multi-track, so everything was in sync already. Now let's just quickly go to see what happened with our truck with unwanted sounds when it's at a sink. I just made it for an example, and this is how it sounds compared with the original step trek. Now let's learn this unsynchronized track instead of a synchronized one. And let's listen. Did you could see it was stronger, they'll collapse, remained untouched, but the ducking occurred at the wrong places. 11. De-click: The DQ click the module helps to reduce or smoother short-term amplitude anomalies which are called Clicks. Clicks going to have different natures. They can distort sound in any way and could be caused by a lot of different reasons. For this example, I prepared a piece of the clean dialogue and the same piece that is affected by a serious amount of clicks, similar to what we can here if we play an old scratch to vanilla on an antique vanilla record player, Let's try to visually compare their spectrograms. This piece, Natalie sounds clean, but also looks clean. This place before the voice looks empty, there is no active background or random noises, and there is no information above 17 kilohertz. This is the same piece with the same spectrogram settings. It looks much busier. We can see this great ash through all the spectrograms and these thin rows are actually the loudest of the clicks. We can also see in this line at around five kilohertz, which is probably a renaissance harmonic. It also presents into clean audio. But here it looks more obvious. We can listen to it with the use of a frequency selection tool. And to hear only the selection we use, not this regular play but this play frequency selection button. And let's listen to this frequency and the clean audio. Yes, here it sounds more annoying and we can get rid of it with the use of different RX tools. But let's focus on this episodes topic which is clicks. Let's go to our D click module. It provides us with four algorithms. Single band works better with the clicks located on the one narrow frequency spectrum. Multiband periodic clicks works well with the regularly repeating clicks on random frequencies. Multiband random clicks meant for random clicks on random frequencies. Low latency, helping when working in real time, especially as a DAW plugin. Of course, there are no rules in using any of these algorithms. And since there is only a few, we can try them all to choose what works the best for us in this particular situation. Here we can adjust how deep it will search for the clicks. So when we hear that the module affects not only the clicks, but also the main audio information. We may want to lower the sensitivity. In frequency skew, we indicate the frequency range priority, whether it's a low frequency range or the high frequency range. Here we regulate how much information we want our x to analyze around the clinic helps when working on clicks with tails. I prefer to start working with some of the repair modules by listening to the output. Note that if I increased sensitivity to the maximum level D click starts to remove too much and we can already hear the actual dialogue instead of clicks only. Let's reduce it and try to switch between the algorithms while listening to the actual audio. I don't worry about people judging me on the things I choose to include in my home. Judging me on the things I choose to include in my home. I don't worry about people judging me on the things I choose to include in my home. I don't worry about people judging me on the things I choose to include in my home. I don't worry about people judging me on the things I choose to include in my home. I don't worry about people judging me on the things I choose to include in my home. I don't worry about people judging me on the things I choose to include in my home. I don't worry about people judging me on the things I choose. I don't worry about people judging me on the things I choose to include in my home. So I think for this example, multiband random clicks works a bit more efficient and cleaner since these are the actually random clicks on random frequencies. I also hear that it's better to leave the sensitivity at around six. Are clicks here are quite high, So I'll keep the frequency skew a bit more to the right. I don't worry about people judging me on the things I choose to include in my home. And I also don't think it's getting better when I widen that click detection area. In fact, the more we widened to click area, the more time it needs to preview or to render. And though the more we widen, the more clicks here are counted, the more here is not always the better. One more thing to consider, the wider the click detection area, the more than main information spreads in a panorama. Let's just drag this slider to the maximum to hear what I'm talking about. I don't worry about people judging me on the things I choose to include in my home. Preview time increase the lat and the voice is annoyingly jumping from left to right. 12. De-clip: The clip is meant to smoother the sine wave parts that we cut while trying to go over 0 decibels. Normally a sine wave has rounded edges, but when it goes beyond the limit, the sinewave cuts at the 0 decibel edge and it's peak becomes square. And the sinewave with square peaks sound distorted. D clip round is these squares which makes audio less distorted or completely undistorted. Let's take a listen to this example. Hello and welcome to the fits William Institute. My name is Annemarie guns and I wanted the lectures here at Prince William. Today. I'm going to be taking you through as part of the diploma in human resources management, the topic of the recruitment and selection process. Let's discover and look at what we're gonna cover in this module today. We can hear distortion on the volume peaks. Most likely clipping occurred at the result of a very high microphone level while recording. But if we look at the picture, we don't see anything that goes over negative three decibels. Where do the clipping come from, then why the edges of the sound wave are squared off? It happens pretty often. The audio was recorded at, at a too high level, clipping occurred and then someone tried to fix it by simply making the audio quieter. But of course it didn't help. Let's open the D clip module. Here we see our clipping picture, but it's not a regular wave form. It's called a histogram. It's sort of an analytic tool to show the sample statistic of the current selection. It shows us the number of samples on a different level. This full length gray line represents the clip. We can zoom in, zoom out if needed. These sliders set the threshold point where the sound wave will start, begin rounded. And what we'd need to do here is to set the threshold level on the point of our clip or anywhere below if we hear we need to. We can also move the threshold level right here in the main area to increase the precision. This suggest button gives a good starting point. I press it and it automatically sets the threshold right on the clip level, which in my case is negative three decibels. The next thing to consider here is if for any reason we have only the positive or only the negative amplitude clipped, we can unlink the threshold and tightened its upper or lower side only. Just like this, the opposite side can remain untouched. But in this example, both sides of the amplitude or clipped, so I'll keep it locked, affect the quality of the processing. I recommend always leaving it as high and lowering the quality only if you're working in real time and your computer can't handle it smoothly. The makeup gain slider will automatically compensate for the volume loss when you tighten the threshold. So if I have negative three decibels right here, I will notice not any volume difference after I apply my D clip, even with the Titus settings while rounding the squared edges, the sine wave amplitude usually becomes a bit more extended. And if there is no headroom between the current peaks and at 0 decibels, the new clips can be occurred while decreasing the previous ones. Keep this post in limiter and enabled to make sure that there will be no new clips generated in the small headroom files. So let's keep the threshold on the clip point and compare the result with the original. All use the Compare button for this. The original. Hello and welcome to the fits William Institute. My name is Annemarie vegans and I'm one of the lectures here outfits William. Today I'm going to be taking you through as part of the diploma in human resources management, the topic of the recruitment and selection process. So let's discover and look at what we're gonna cover in this module today. And the result. Hello and welcome to the fits William Institute. My name is Annemarie vegans and I'm one of the lectures here at fits William. Today I'm going to be taking you through as part of the diploma in human resources management, the topic of the recruitment and selection process. So let's discover and look at what we're gonna cover in this module today. Absolutely amazing. Let's render now. See the edges of the sound wave. Now. There are all smoothly rounded. Now. Everything sounds organic without any distortion. 13. De-crackle: With the use of the crackle module, we can partially or completely get rid of the group of dense clicks called crackles. This is one of the modules I rarely use as the main tool and more often as an element of the repairmen chain while working with difficult audio recordings. D crackle has quite a simple set of control elements. Here we choose the combination of the CPU usage and the processing quality. Here, how heavy the processing we want to be. And the amplitude skew lets us choose the approximate location of the crackles on the amplitude of the main audio. The more to the right means the crackles are located on the top edges of the amplitude similar to the clips. And accordingly, the more to the left means they're crackles are at the low range of the amplitude. Let's listen to the original audio. I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. We can hear a sort of distortion behind the voice. Though the crackle is a pretty simple module. It also requires a bit more resources than most of the other RX modules. So it's pretty hard to hear the result in real time when the high-quality processing is chosen. This is while use the Compare button, it allows me to create an unlimited amount of previews while with different settings. Let's say the first file to compare will include only the default settings. The second I'll go for the maximum strength. Then I'll play with the amplitude skew 3510, negative three, negative five, negative ten. Then we'll listen to them one-by-one, being able to switch between them anytime to insert in the year. The difference, I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. I believe decluttering is a powerful part of the process of simplifying your life and creating more time and space for what you love. I like this one the most. The full stretch and the amplitude skew on three. Although it didn't clean cracks completely, it definitely cut all the high-end mid-range frequencies from them, making them less noticeable and less annoying. But as I said, the crackle works best when combining it with other modules, we may achieve more results here if we use D click first, for example, and then polish the outcome with the crackle already. 14. De-ess: The ester is a very well-known tool that is used not only in audio repair, but in vocal music mixing and mastering and any kind of dialogue editing, what's so special about it, this tool reduces the harshness of the siblings sounds, making the listening process more pleasant and comfortable. Let's listen to the example I prepared for this episode. I want to note that if you're using some sort of professional monitoring speakers or headphones, this piece may not sound as sharp as it sounds on a regular average price domestic device. But as professionals, we have to deliver an ultimate product that will sound the same comfortable on any audio device. I've been so busy with school and volunteering at the animal shelter. I just got confused. So I can note a few of the sharpest moments here. So school volunteering, shelter, just confused. If you listen to this example loud enough. These letters are just cutting through the years. Let's open the DS module. I can say that the way ds or works is similar to multiband compressor or even the dynamic equalizer. But unlike most of the compressors and equalizers, ds or uses a smart algorithm that searches and affects only the siblings, leaving all the other sounds untouched. There are the classic and the spectral modes available for us here. And roughly speaking, the classic is a simple go-to mode that will affect all the found siblings in one setting. And it's quite enough in most cases when we aren't going to re-listen to the phrase 50 times to find the smallest issues. But if we need to achieve the best possible result, we'd probably want to choose the spectral mode, which provides very flexible surgical and gentle DSM settings, something we need when working on a top-notch vocal production, for example. But let's come back to the classic mode and let's see how everything works here. There is a threshold slider, and by default, it said on negative 12 decibels. So only those siblings that are louder than negative 12 decibels will be processed with this module. This is the loudness scale related to the current level of the audio. But if we want to use the full-scale loudness measurement, we take this checkbox and we see that the measurement here changes from decibels to dBFS. Here we will see the current level of the sibilant and here the level of its reduction. This is the frequency threshold. Right now it will search an effect only the siblings above 2500 hertz. Faster and slower the speed modes of the DS, his reaction towards detected siblings. Let's try changing the settings while previewing and the classic mode. I've been so busy with school and volunteering at the animal shelter. I just got confused. I've been so busy with school and volunteering at the animal shelter. I just got confused. I've been so busy with school and volunteering at the animal shelter. I just got confused. I've been so busy with school and volunteering at the animal shelter. What can we say about it? Of course, the more I typed in the threshold, the quieter becomes the audio. Because the S or finds more siblings and tightens them harder. And of course, when I switch to the absolute measurement, I need to tighten the threshold more because simply in the Audio related scale, this is my 0 decibel edge and negative 12 decibels is somewhere deep. But in the full-scale, this is a 0 decibel edge. And here is a negative 12 decibel line, which crosses less than half of the average amplitude of my audio. So accordingly, I need to tighten the threshold more in the absolute mode for the DS or to catch more of those siblings. I've been so busy with school and volunteering at the animal shelter. I just got confused. I've been so busy with school and volunteering at the animal shelter. I just got confused. Moving the cutoff slider doesn't make much difference right now because it's a pretty high voice and most of the siblings in it that classic mode can detect are located above 8 thousand hertz. Whether I pull it all the way down or up, It's still suppresses the siblings above eight kilohertz because there is nothing classic mode that can find below a thousand. Of course, you should be careful with this slider when working on the voices in its lowest registries or when working in the spectral mode. I've been so busy with school and volunteering at the animal shelter. I just got confused. I've been so busy with school and volunteering at the animal shelter. I just got confused. I also don't hear much difference between fast and slow boards in this case. And overall, I'm happy with the result. Ds are performed very well here, and even the default settings made this dialogue much softer and it doesn't slice my ears when I'm referencing in my ear buds. But even if we retrieved a great result, can we still improve it? Let's get everything back to default and use the compare function. Now I have here the original dialogue and the same dialogue processed with the default settings of the classic mode. By the way, we can now visually compare them and see these peaks are lowered a lot after the processing. Let's switch to the spectral mode. Now, here we have two more sliders. Spectral shaping is how much the sibilant will be flattened and the spectral tilt is the way it will be flattened. In other words, let's represent our siblings in ADSR form. So the first slider regulates how flat it'll be in this area from the attack to the decay. And the second slider regulates the type of decay siblings will have. The decay type here is related to three noise types. So the more we move the slider towards white noise, the more high frequencies will be preserved, making the sound a bit brighter. And the other way around, The more towards brown noise, the less high frequencies that K will have and the darker sound we will get. Let's listen now and let's maximize the spectral shaping so we can better hear the impact of the spectral tilt. I've been so busy with school and volunteering at the animal shelter. I just got confused. I've been so busy with school and volunteering at the animal shelter. I just got confused. I can hear the sound is more sort of open in white noise in comparison with brown noise. Now let's reset everything to default and make one more file to compare. I press compare and we see one more position in this list. The first one is the original, the second one is the icing and the classic mode with the default settings. And the third one is decreasing in the spectral mode and also the default settings. Let's listen to them one-by-one. I've been so busy with school and volunteering at the animal shelter. I just got confused. I've been so busy with school and volunteering at the animal shelter. I just got confused. I've been so busy with school and volunteering at the animal shelter. I just got confused. Yes. The classic mode provides heavier siblings suppression, leaving no place for ear irritation. And the spectral mode gives a gentle touch, making this piece sound more natural and transparent. 15. De-hum: The hum is meant to deal with the static low-frequency noise that some electrical devices generate as a result of bad grounding, weak electrical contacts, poorly isolated wires, increased electrical resistance and so on. In this episode, we're going to get down to 60 hertz. So I suggest you use good quality speakers or headphones because there's no chance to hear sub frequencies on your laptop or cell phone speakers. I prepare two examples. We will be checking the DOM module on. Here. I've got an acoustic guitar and piano with a sub frequency hum. And the second tab is a female voice with multiple harmonic humming. Let's start with the first example. If you listen to it loud enough, you could recognize the tonal bass note going through the audio. Since it's constant in quite recognizable, we can easily detect it's visually. Here. It is bright like array of sun. And we want to get more control over it. We can change our frequency scale to log or even to extended log. And let's also improve the resolution of the visualization. Right-click View spectrogram settings and change the spectrogram type to the adaptively sparse. It takes a few is to recalculate. And here we go. If we zoom, will see that the heart of this line is exactly 60 hertz. 60 hertz is the North American electricity lines standard, but most of the countries use 50 hertz standard. So if this jam was recorded somewhere in Europe, we would see this bright line in front of 50 hertz. Let's take a frequency selection tool. Select it, and listen. Nothing but a pure base. In this example, neither guitar nor piano is reaching low. So we could have just press Delete now and forgotten about this humming once and for all. But let's imagine we've got some information there that we want to preserve. Let's finally open D hub, which may look like a spaceship control panel when you open it for the first time. But though it has a lot of settings that are quite simple, the module has two main work modes, filter types, the dynamic, which is designed for the hummingbird, consisting of multiple harmonics. And the static, which is meant to deal with the humming with a small number of harmonics. Since our first example, it has minimum harmonics in it. We're choosing the static type. We could have also gotten rid of it and dynamic mode, but let's follow the developers recommendations. In this window, we will be seeing our real-time audio gram. And with the use of these cuts, we decrease the loudness of certain harmonics. The process is similar to digital equalizers, but the D HM module is more harmonic oriented tool unlike the regular equalizer. So how does it work in general? We choose the first, the fundamental harmonic of the humming. We can use both this crossline cursor and this slider. We choose its level and the width of the cut, the precision, the sharpness of the cut calls q, just like the equalizers. And accordingly, the narrower this Q, the more precise and accurate the cut is. But of course, this is not always what we need. After we found the fundamental harmonic, we can add up to 16 more cuts relative to the additional harmonics built from the fundamental one. Like this. Let's say my fundamental harmonic is 100 hertz. The sound wave is dividing equally. And accordingly, the second harmonic is 200 hertz and so on. And naturally every next harmonic has less amplitude, which means they are fading towards the base, the fundamental harmonic. This is why we have this slope slider here, which lets us reduce the level of every next cut. These harmonics here will be quieter, which means we will need less reduction on them. Here are the options for how we want to link the depth of the cuts. This may help to fight the hummingbird buzzing with unequal harmonic structure. We can also enable and set the high-pass and low-pass filter. The Learn button here helps when we have an example of the single humming without useful information or when the humming is very obvious. We just selected and learn here how my humming looks like. Here's its peak. And accordingly, RX automatically detected the first codon there. And before we actually hear it and start doing something, I'll mention the last option here, the adaptive mode. Let's this module adjusted settings in real-time while previewing the audio. And it's the opposite of learn. When we learn we adjust the settings we get when we learned the piece to the whole audio. Linear phase filters. Simply saying this option increases precision together with the latency. So if you feel an uncomfortable delay, try to turn it off. The last year is the filter DC offset. If it's enabled, the module try to fix the amplitude displacement from 0 if this problem was detected. Let's finally get into practice. Here's before. Here is after. Awesome. I can also use the Learn button since I have a pure humming in the beginning before the music. And it suggests the same 60 hertz frequency. Let's switch to the dynamic mode will we will be working on the second example. Here we only have three sliders, which are the sensitivity it thoroughly searching for the hub bands, the number of cuts will be created. And the filter is the width of the cuts. It's barely noticeable visually, but the impact it creates is noticeable a lot. This gate is just the processing zone selection. This part will be processed with the module and this gray zone will remain untouched. And again, please make sure you have enough volume on your headphones or speakers because the low frequencies require more gain for our ears to hear them. Let's listen. Over the next two weeks, I was obsessed with any package that came to the house. Anything that held the shape of my jewelry box was torn open. This often led me to getting stern stairs from my parents. However, I didn't mind. I needed to figure out where Gideon place my jewelry box. Here. We also have a space before the dialogue where we can hear only the noise. This one sounds more distorted, more like a buzz. If we took a look at the spectrogram, we can see the nature of the sound. It has multiple harmonics. Here they are. The first one starts below ten hertz and then 50, One 100, and so on. Let's try to learn this part where we have nothing but noise. Over the next two weeks, I was obsessed with any package that came to the house. Anything that held the shape of my jewelry box was torn open. This often led me to getting stern stairs from my parents. However, I didn't mind. I needed to figure out where Gideon place my jewelry box. The noise was removed a lot, but let's try to add more bands, widen them by lessening this value. Over the next two weeks, I was obsessed with any package that came to the house. Anything that held the shape of my jewelry box was torn open. This often led me to getting stern stairs from my parents. However, I didn't mind. I needed to figure out where Gideon place my jewelry box. Now we remove the noise even more without causing much damage to the voice. 16. De-plosive: The plosive is very simple, but at the same time a very important module. It helps to smooth in the harsh plosive sounds. Strong and abrupt air flows that hit the microphone diaphragm when pronouncing certain letters pretty close to the microphone. In the English language, the letter P is the leader in the amount of air pressure we put into it. And if the speaker or the singer didn't use the pop filter while recording, we may have a bunch of unpleasant low-frequency splashes. Let's listen to the example. Painful period. Prison, paid a plank poison. Each new letter is just punching the ears. Let's go to the dead loss IV and quickly do something about it. Why quickly? Because the module only has three simple sliders. Here we set how sensitive it'll search for this plosives. Here, how hard it will suppress them. And here, the maximum frequency until we want the explosive to be processed. If the plosives we're dealing with are mostly located on the lower end, the lower this value to preserve any useful information above. The other way around. If it's pretty high frequency plosives, we increase this value. Otherwise, it's just not gonna work. Let's start with the default settings. Painful, period. Prison paid a plank poison. Maximize the strength, increase the sensitivity. Painful period. Prison paid a plank poison. And if I increase the frequency limit, painful period, prison plank, always, we get more reduction in exchange for us sort of ducking, which is not what we want. I better keep the frequency limit at around 200. The original painful period. Prison paid a plank poison. The result painful period, prison paid a plank poison. It definitely got much better. Though. We still hear some punch above 200 hertz in almost all the words. Fair to say those words were intentionally recorded to showcase plosives. And normally plosives aren't located that close to each other and not a lot of them are so strong. So the module did a pretty good job here. 17. De-reverb: D reverb is designed to remove reflections from the original signal. I assume if you're watching this course, you have at least a basic knowledge about the nature of the sound and you'll understand what reverberation is. But for those who are taking their first steps in audio engineering, I can shortly say that reverberation is the reflection of the sound waves from surrounding surfaces. Every time we speak, clap, sweep the floor, brush the teeth, you name it. Every time. The waves of every single sound spreading around, hitting the walls, floors, ceiling table, mirror, turning around and running around until completely fading out. The recording should not contain any room sound in it. But unless the recording was made in a very well acoustically treated room, we will have a reverberation mixed in this original sound. I prepared two examples for this episode. The first one is the acoustic guitar recorded in the lecture room. The second one is the kick and snare of the drum kit recorded in the average size live music club. Let's go to the D referred module and discuss its controls. Let's press learn and play a little so we can see the information in these two windows. The grayscale is the signal of the original unprocessed audio. The white scale is how the signal looks after the processing. And this orange scale is the difference between the gray and white between unprocessed and process signals. This is what RX estimates as reverberation. The slider adjusts how hard the reverberation will be suppressed. In general, here we set the reduction strength of particularly frequency ranges. The more up, the stronger the reduction is. The tail length is how long? If we think the reverberation lasts, It's an approximate value and we may get an idea of how long it is by listening to the end of any audio phrase. Like here, for example. We are already familiar with the artifacts smoothing slider. The more the value closer to 0, the higher the risk to get artifacts similar to the low resolution mp3 sound. And the closer it is to the maximum, the more kind of underwater sound we're a risk getting. This function helps us keep the loudness on the same level after the reverb reduction. Although sometimes it may create sort of a side chain fields, so use it wisely. Let's learn again and listen. We can also solo one of these frequency ranges and listened to them separately while previewing. This helps to detect if there is more or less reverberation in this particular frequency range. These orange peaks here on top are the amount of the signal that was removed from this frequency range. And when we see the same peaks at the bottom, this is the signal enhancing. It appears when we enhance the dry signal or if we drop the reduction slider below 0, it starts increasing the reverberations. Instead of decreasing. It does a good job here. And let's see what we can do with the second example. Let's learn and listen. Now it obviously cuts too much high-end from the snare, while the kick lost a little of the power with the low frequencies, I think I'm going to put a high frequency reduction to 0 and also minimize the high mid reduction like this. What we can achieve here is only less than the reverb lesson, the room size where the drums were recorded. We cannot make a completely dry without causing noticeable harm. We could achieve a bit more flexibility if we had this kick and snare separately. And in this example, it's very easy to do since we don't have any symbols that would sustain through this kick and snare. But in my opinion, when we have this straight kick and snare play, we can make them dry without much harm by simply cutting the tail. Let me demonstrate we didn't cover it yet, but I already mentioned this very handy module call fade. What I'm going to do is copy the first kick and snare hit. I'll select a little more. I open file, select New from clipboard. Here it is. Choose fade out. The mode of how it will fade. I'll choose the last one. Now I select the kick and let's compare the difference. Here is the original. And this is how it sounds without a tail. Pretty cool. The kick still has its power but sounds very dry. So that's it. Let's compare the original kick and snare. And the kick and snare with faded tails. Actually, we can fade the tails in any audio editor and any DAW. But I just demonstrated how we can do this in R x and how in general, this procedure helps to remove reverb in the single samples. But of course this method isn't going to work and complete trump part or any part with more than one sound at a time. 18. De-rustle: The next module is called D Russell. The Russell doesn't have a specific nature or characteristics or even exact description. It's a pretty huge noise category. This is why D Russell is more of a team player rather than a single fighter when it comes to multi frequency dynamic type of noise. Another thing to keep in mind when working with a wide frequency range dense type of noise is that most likely we won't be able to get rid of it completely without damaging the main information. There is always a balance between the amount of Russell we want to remove and the amount of damage we can accept. The module only has three control in which our reduction strength, just like everywhere before it adjusts the amount of processing energy. Ambience preservation helps differentiate the useful ambience from the unwanted Russell. And here we choose the processing algorithm. Channel independent r x will scan and process every channel individually joined channel. The module processes all the channel and mono mode, advanced joint channel offering the highest quality but at the price of longer preview time. And in the case of the Russell interrupted preview depending on your computer horsepower. But let's get to the example. Here is a short phrase with a typical clothes Russell. There's got to be an easier way than working all hours for a Pitney. The lavalier microphone often catches these noises when the person is moving, while speaking. Maybe only moving his hand, but wearing some kind of not very tight clothes will be enough for the mic to catch the Russell. Let's preview and I'll be mowing the reduction strength to the maximum because I'm pretty sure this is what we're going to need here. The ambiance preservation here, it doesn't make any difference because there is no ambient in recording. I'll create compare previews for each of these algorithms. There's got to be an easier way than working all hours for a Pitney. There's got to be an easier way than working all hours for a Pitney. There's got to be an easier way than working all hours for a bit. And let's render. And we see that the spectrogram in general became cleaner from this gray dust switches are Russell. We can switch between the original audio and the processing I just applied right here in the History Window. Also, the wave form amplitude became smaller. In some cases, we may render the audio twice and doubled this result. In this case, the audio will lose a lot of its quality. And the second D Russell. So to improve this result, we'd better use the other modules such as voice de-noise and so on. What I don't want to do this now because as I mentioned in the beginning in this chapter, one episode is dedicated to one RX module only. I'm trying to keep the clarity so I don't want to combine them in one video, but this is what we're going to do in the last episode. We will be trying to solve different problems with different module combinations. And of course, this rustled recording is also available for you just like all the others. You can try to improve it with just the use of the additional modules. 19. De-wind: The wind is one of the main enemies of the outside world recordings, depending on the microphone type, the strength of the blow and airflow interaction angle, we can get various types of noise from the slight whistle to the heavy low-end rumble. D wind is more focused on the lower part of the frequency range, and it's designed to fight the light and medium strength impact of the wind. The module only has a few control elements. Here we choose the strength of the render reduction here until what frequency it'll look like for signs of the wind. If it's a 1000 here, nothing above 1000 hertz will be affected by this module. The wind also knows how to recreate voice fundamental harmonics that has been lost because of the wind or the wind reduction. And with the use of this lighter, we adjust the amount of these harmonics that will be synthesized. The artifacts smoothing slider is also very familiar to us already. It helps to lessen the consequences of the raw processing with the risk to get a bit dull sound. So the artifacts moving is often to finding a balance. I've got two windy recordings here. The first one is a slight surrounding Wessel. If you keep the doors open or one door like I've got there, if there is some rain, it comes in a bit sideways, it's gonna go straight through that mesh. And the second one is a hard punchy and wind blow your hair doing and you can tell, not too good. The wind takes a while to build a preview, so I'll use the Compare button for us to not be interrupted while previewing. Let's start from the default settings. It takes some time to build a comparison option. And here we go. And note that the low-frequency part has changed, but the upper part looks the same as I mentioned in the beginning. And it's also how the isotope describes its tool that the wind is more low-end focused module. And there is something we should understand when working in RX is since the noises don't follow some sort of strict technical specifications, there cannot be a single ultimate tool for it. And if we need to reduce the wind present in the recording, it doesn't mean we can only use the D1 module for it. Maybe in this case it's worth trying. D Russell, for example, spectral de-noise will do a better job. Next, modules compliments each other and we should trust our ears and our experience. But this video is about the DEA window me. So let's compare the original with the default, the wind settings. If you keep the doors open or one door like I've got there, if there is some rain, it comes in a bit sideways, it's going to go straight through that mesh. Keep the doors open or one door like I've got there. If there is some rain, it comes in a bit sideways, it's gonna go straight through that mesh. It's just the same as we see in the spectrogram. The changes are only below a thousand because we have 1000 hertz here. So nothing above was affected. The wind became a little weaker, but they'll voice lost some of its low energy as well. Let's build a few more comparison options with the highest reduction level. The first one will be with a maximized crossover frequency and fundamental recovery. And the second one with a maximized crossover frequency, but minimized fundamental recovery. Keep the doors open or one door like I've got there. If there is some rain, it comes in a bit sideways, it's gonna go straight through that mesh. If you keep the doors open or one door like I've got there, if there is some writing, it comes in a bit sideways, it's going to go straight through that mesh. I honestly don't hear much difference between the two of them, but the production level is definitely much more noticeable than what we got while previewed with the default settings. If you keep the doors open or one door like I've got there, if there is some rain, it comes in a bit sideways, it's gonna go straight through that mesh. If you keep the doors open or one door like I've got there, if there is some writing comes in a bit sideways, it's gonna go straight through that mesh. In general, the wind became less destructive. They'll, they'll voice has lost a bit of its character. So the result here is pretty questionable for now. And again, we might improve the result of cooperating with the other modules. What we're going to do next is try applying the same compare settings to the second audio. We start from the default settings. The second is where the maximum parameters and the third one is with the minimized fundamental recovery, but with the average artifacts smoothing. Let's listen. It's freezing the wind in your hair doing and you can tell not to guide. Its freezing the wind is how's your hair doing it? You can tell, not too good. It's freezing, the wind is outrageous. How's your hair doing? And you can tell, not too good. It's freezing the wind in your hair doing it. Because you can tell not too good. Of course, the default settings are lacking the reduction level here since it's very heavy wind. And the difference between the second one and the third one's still isn't audible for me, but the difference is visual. Here is the generated fundamental harmonic. This is the variant with the maximized fundamental recovery. In the variant with the minimized fundamental recovery, this harmonic is missing. If I leave these harmonics, they won't make much difference now, but they might make a difference if will go for additional processing with other tools. 20. Deconstruct: In my opinion, deconstruct his one of the most powerful tools in isotope RX. It separates the audio signal into the tonal elements, noise and transients, and lets us control the amount of each component in the recording. Here are the controls for each of them, the level of the tonal components, that noise level, and the transient game. It's turned off by default. Using the tonal, noisy, balanced slider, we regulate how thoroughly to deconstruct the module will look for the difference between the total elements and the noise. The more to the left, the lesser the difference and accordingly, the lesser impact we get and the more to the right, the more separation will take place. The artifacts smoothing helps us to lessen the low-quality field that we get after applying type processing to the audio. And for the example we have here already familiar to us seashore podcast recording. I actually find it much, much easier to have social interaction here. It's the same recording as we had many ambience match module review, but the different part to it in that episode, we just needed to make two different pieces of audio sound alike. But in this episode, we will try to lessen the ambiance since it's very strong and quite destructive from the dialogue. Why lesson? Because if we try to clean it up completely, we will lose a lot of useful information. Even though RX plugins do a great job in audio restoration, they don't do miracles. Not yet. We can tell that it's a very noisy recording by looking at its spectrogram, the brightest parts or the voice, we can loop them and listen. These are the loudest parts of the words. And this orange background is the strong gambiense. We can also say that the audio is cut at a very high end. Let's switch to the linear frequency scale. Now when all the frequency ranges are represented equally, it's easier to see that it's been cut it around 16 K, which is a sign of a medium quality MP3. But now this file, just like all the other audio for this course, isn't a WAV format because I converted it simply because when you load mp3, RX starts converting it into wave, which may take awhile. Alright, so since we have both tonal and noisy gain reduction sliders set in decibels, the default 0 value won't make any difference now because it's related to the audio signal, whereas 0 decibels is their current level. I think I'll start creating comparison options from the most radical settings here and then easing them a little by little. But I won't maximize the total gain because the voice here is not too quiet. If I maximize it, I'll get nothing but extreme clipping. So three decibels here will be more than enough. Let's see what's gonna happen if we reduce all the detected noise to the infinite level. Let's leave the other parameters by default, compare. Find it much, much easier than social interaction here. Just like I mentioned at the beginning, if we try to clean it up completely, we will lose a lot of useful information. This is exactly what happened here when I decided to minimize the noise level. Of course, this is way too much, but this is a good chance to check the impact of the artifacts smoothing. Right now it was on its average value. But let's try to maximize it and compare. Just listen to both of them. I actually find it much, much easier than social interaction here. I actually find it much, much easier to have social interaction here. What happened now is a lot of this high-frequency junk disappeared in the second example, when we maximize the artifacts smoothing, it's great. The voice is still very dull in the words in general are less recognizable than in the original, which is not what we need. We need to find good balance between the amount of noise we want to remove and the acceptable damage to the voice. Let's rise the noisy gained to, let's say negative 20 decibels and create one more comparison option. A very handy feature here is the view settings option. When we create a lot of these comparison options, we can easily forget what settings each of them contains. And when we select one of the comparisons here, we can click View settings to see what we tweaked here to get this sound. Let's listen. I actually find it much, much easier to have social interaction here. This one sounds better to me. We brought back a certain level of the ambience, but also restored very important voice information that we lost when we minimize the noise. Let's try the transient separation. The transients are the short peaks at the beginning of the waveform. And if we raise them now it'll sharpen the audio. Let's listen. I actually find it much, much easier to have social interaction via. It doesn't sound like it improved the voice but only raised all the minor calyx we were paying attention to before. But we can also reduce them and let's drop it a lot to, let's say negative 30. I actually find it much, much easier to have social interaction here. Yes, it definitely cleaned the audio from the clicks. I like the result, and let's compare it with the original. I actually find it much, much easier to have social interaction here. I actually find it much, much easier to have social interaction here. 21. Dialogue Contour: Dialogue contour is a very interesting and useful tool that lets us manipulate intonation of the words and sentences. How does it work? Have you ever noticed how a question differs from a statement? Mostly by pitching. Roughly speaking, the question word rises and tone towards the end. The statement word is not changing or even goes down in pitch. The same happens where a person hasn't finished the sentence yet. The end of the last word before the pause will be higher than if the finishes the sentence. The module has simple but very flexible controls over the pitch. Let's discuss them while training on the exemple. Let's listen, please become overprotective again. So we can hear that starting with protective to the pitches rising. And it's not bad, but it creates a sort of suspense field and kinda wants us to wait further continuation. So to make the phrase sound more final, we need to lower its last part. How to do this? For starters, let's select the part we want to pitch. Now let's open the dialogue contour. On top we see the waveform of our selection. We can't change it though, we don't need to. This just shows us the current level of the selection and it will change once we apply some changes. This spectrogram is the main working area of the module. This is the pitch curve. It's straight by default, but if we bend it up or down, it will accordingly raise or lower the pitch. We can create up to 25 dots and change the curve between them in any way. Let's reset everything back to our selection. We need to lower the last part of the sentence to make it sound more finalized. We can see the number of semitones while moving the cursor over this area. Let's lower the part on, let's say 3.5 semi-tones. We're not going to lower it all on 3.5 semitones from the beginning. If we do, it will create a huge gap with the first part of the sentence and it will sound synthetic, weird and bad in general, what do we want to do here is create a gradual pitch lowering is simply like this. The first dot remains the same, and the last dot is 3.5 semitones down from the original. Let's compare protective again. Protective again, protective again, protective again. Now the whole phrase, please become overprotective again. Please become overprotective again. In general, it provided me with the result I wanted. But now it sounds a bit robotic at the pitch part. This is a common side effect of using any kind of pitch tool like Melodyne, AUTO-TUNE, and so on. So what I'm gonna do now is try to make this artifact it less noticeable by tweaking this curve. When we have more than two dots here, we can smoother these sharp angles by dragging this slider. Let's try this one. Please become overprotective again. Yes, this sounds much better to me. Let's remove these thoughts by right-clicking, or we can just double-click on the dock to return it to the default position. Before we jump to the second example, I wanted to discuss the last two sliders here. For men scaling formats are frequency peaks in the spectrum which have a high degree of energy. They are especially prominent in vowels. Each format corresponds to a renaissance in the vocal tract and simple words. Because of the formats we can tell the voice of the young person from the voice of the old person, even with the same pitch. This parameter works only if we have the curve changed because there's nothing to scale when there is no pitching. So simply saying, the more this slider to the left, the more formats are staying in the original tune while the other elements of the signal are pitched, it creates a sort of distortion, but at the same time, it gives us special character to the sound. Please become adopted, we're done. We can easily recognize it. Producers are massively using it in the future. Bass and lots of other modern musical genres. Kids like using this pitching method in different entertaining Tiktok videos, YouTube shorts and so on. And the more we drag this slider to the right, the more formats are following their pitch and it creates a more clear, more natural transpose. Please become overprotective again. Please become overprotective again. In many cases, using format scaling also helps to get rid of at least a minimizing the pitching artifacts. And the pitch offset is the complete pitching of the whole selection. I mean, if we drag this slider to the very left, it'll be just the same if we pull this curve completely to the bottom. Let's compare. Please become overprotective again. Please become overprotective again. Sounds absolutely the same. But if we do this at the same time, if we pull down the curve and drag the slider to the left, this will double up the pitch. Please become overprotective again. And accordingly, we can compensate, for example, if the curve is on the bottom, but we drag this slider to the very right. It will now sound the original. Because here we offset to negative six semitones and here we added six semitones, which give us 0 pitch. Let's compare. Please become overprotective again. And there are no rules at all, no steps or methods in dialogue contour. We just listened and tweak until we find what we need. We can quickly get a great starting point by using presets for the beginning of the word ending offset. And it's all very fast and simple. The module doesn't require many resources. It works perfectly in real-time. The next example is very short, but it's a good chance to explore on little more of the dialogue contour opportunities. Thank you. For example, if we cut like something between Frank and you will change the destination of the phrase. The original sounds more general. Thank you. And now it sounds like the person is singing out loud to someone who he thinks. Thank you. Maybe in a bit ironic way. But what matters is the voice expression has changed. And of course we need some contexts to find out if the new expression fits. It. Feels like sculpturing that is, don't be afraid to experiment to get a creative outcome. 22. Dialogue De-reverb: Dialogue D reverb is designed to remove or less than the reflections from the audio signal. We already covered the general deep reverb RX module, which is a strong and ultimate tool for rebirth suppression. And as we already understand from the name, the dialogue D revert is a module with a narrower task. It's machine-learning algorithm is programmed to trace and remove the reflections of the human voice. The dialogue D reverb has a much more simplified interface than his big brother. We can also compare it visually. It also does not have a Learn function because the analysis and dialogue, the reverb is adaptive. In general, we have already seen these controls multiple times in previous modules. So they are already very well familiar to us. This is how strong we want the reduction to be. It goes from top to bottom. If the slider is at the very top, there will be no reduction at all. This very bottom minus infinity value means the module will suppress the rubber innovation as much as possible. This is how thoroughly the module will look for reverberation. And here we'll adjust how hard it will separate the reverberations from the ambiance. If we have some important ambience, we keep the slider more to the right. And if there is no ambience, we keep the slider to the very left. There are three separation algorithms available in the dialogue D reverb, and we can sort them by processing time and quality balance. Channel independent is processing all the channels independently. It is the fastest algorithm and at the same time, it provides lesser quality compared to the following two. In joint channel, all the channels are processed as a whole. As the result, we have better quality, but also slower processing compared to the channel independent algorithm. The advanced joined channel provides us with the best quality and longest processing time. Everything is quite fair. Let's quickly check on how it works on the example I prepared for this episode. I don't think it's necessary to try every algorithm and every position of every slider here. Because though it's very smart and powerful, it's very simple and understandable. Besides this audio example, just like all the others is available for your training. Let's start with the default settings and note that by default it's already at maximum reduction level. But don't be afraid because dialogue D reverb is much more gentle than the General de reverb module. So even it's on the maximum in the cases with the small and medium-sized reverbs, a won't kill the original sound. Let's listen. I took it into my room, closed my massive doors and blasted on my speakers. Took it into my room, closed my massive doors and blasted on my speakers. The first part is great, but later in the loudest moments, the reverb is still pretty audible. So let's just increase the sensitivity to around seven. Took it into my room, closed my massive doors and blasted on my speakers. Now it's way much better. And let's maximize it for comparison. I took it into my room, closed my massive doors and blast it on my speakers. Now it's super dry and I haven't noticed any critical quality loss of the voice, which is incredible. If we long for more surgical reverb cleaning, it's worth rendering twice the parts with the most obvious reverberation and carefully going through the audio with a DSLR or even high cut EQ. The clean those reverb leftovers that are hiding on frequency peaks. 23. Dialogue Isolate: Dialogue Isolate is one of the strongest noise reduction modules and RX, it's designed to detect any kind of human speech in the recording and separated from the background noise. The interface of the module is very simple. Here we adjust the gain amount of the dialogue. We can make it louder or quieter towards the ambiance or any other background noise. When 0, the dialogue level remains original. And accordingly the level of the background noise, how much we want to reduce it. Or if we drag it above 0, we will even increase it if we need. This is how accurate the module would differentiate the dialogue from the ambiance. And this is how much ambience we want to keep. Why do we need the ambiance preservation if we can adjust the amount with the noise gain slider, because the noise game removed and not only the ambience, but any kind of non-speech sounds. And if we increase the ambiance preservation, it will increase the amount of only the clean steady ambience without random noises. But in this example we have pretty steady background noise. So if we need to keep some ambiance, we can use both sliders for it and the result will be pretty much the same. And again, it is a different part of already familiar to others recording the beach line. I just find that a good example of a dialogue with the massive background noise. I admire. They have lots of people who will just come up to me and say, I'm actually quite shy. There are two processing algorithms available in dialogue Isolate. Good, which is a bit faster but less precise, and best which is a bit slower and provides the best quality possible. Note the dialogue Isolate is very effective, but also of pretty slow modules. So even if you choose a good algorithm, get yourself ready to wait for a little every time you're previewing or rendering. The voice here is jumping on volume. These peaks are quite high, so there's no point in increasing the dialogue game. We will also keep the noise gain and the ambiance preservation at their minimum, which is by default, the sensitivity in the middle also by default. Let's create a preview of this piece which the good and best algorithms. It will take a while. So I'll speed up the video. We can even tell by spectrogram, the audio became a lot cleaner. And yeah, if we listen, I admire. They have lots of people who will just come up to me and say, I'm actually quite shy. I admire. They have lots of people who will just come up to me and say, Hey, I'm actually quite shy. I admire. They have lots of people who will just come up to me and say, I'm actually quite shy, though it's still requires some work to do. The result is also very cool. The difference with the originalist significant. The difference between these two algorithms is barely audible. Though the spectrogram of the best algorithm is a bit cleaner than the good one. And lastly, let's try to maximize and also minimize the sensitivity. According to the official manual, higher values separate less noise from the dialogue. I admire. They have lots of people who will just come up to me and say, I'm actually quite shy. I admire. They have lots of people who will just come up to me and say, thanks. I'm actually quite shy. And it's absolutely right. 24. Guitar De-noise: Guitar denoise is a fairly new module. It was first introduced in the RX eight and is meant to deal with the three most common types of guitar noises, which are electric guitar, buzzing strings, squeak, pick attack noise. So the module is divided into three sections that are dedicated to each of these noise types. I also prepared different examples including all of these problems. Let's start from the first amp section. First, if we want to use only this tool, we'd probably want to turn off the other two to avoid unwanted processing. We press here and it becomes inactive. To make the amp section useful, we need to show it in the example of pure Buzz, we felt the main signal. In this case, these pauses between riffs are the best places to learn. I select one and press learn. Now I want to notice that the buzz in this recording is very strong. And even if we maximize the sensitivity slider, the module will be only able to reduce it but not remover completely. The sensitivity slider here performs the gain reduction. The more up, the more bizarre will decrease and the resolution slider is how many harmonics we want to remove. Let's listen. While I don't hear any difference from changing the position of the resolution slider. And as I said, the sensitivity even being maximized only slightly reduces this buzz. Yes, it became less annoying but still very noticeable. For better results here we could try to also use the hum, spectral de-noise and so on. But this episode is only about the guitar denoise. Let's move on to the second section of this module. I'll turn off the AB and turn on the Squeak. Let's switch to the second tab and listen. This is the typical noise that occurs almost whenever a guitarist quickly release his fingers from the strings while playing. This happens both on acoustic and electric guitars. And the newer the guitar strings, The louder and more obvious these squeaks are. Please note that the squeaks are an important part of the instrumental character, which is why we don't have to remove them completely. What do we want to achieve in the vast majority of cases is to quiet, intend them are little and make them less sharp. In this case, the sensitivity slider is responsible for the accuracy of the squeaked detection. And the reduction is how hard we want to suppress them. Here we can also switch between the short and the long modes. In short mode, RX will look for the squeaks up to 200 milliseconds and in Longboat up to one hundred, ten hundred milliseconds. In theory, the faster the temple, the faster the guitarist is changing the position of his fingers toward the guitar frets. And accordingly, the shorter or the squeaks. And the opposite with this lower piece. But in practice, it depends a lot on the guitarists playing technique and his experience level. And if we enable this gear icon, we will be hearing the removed squeak sold me. It's always a good idea to monitor in the output mode to make sure we don't remove any useful information from the signal. Let's listen. I think we can increase both sliders up to around five. And it's great. Squeaks are still well audible, but they don't drill anyone's ears. We can see them lowered a lot in the spectrogram. Let's try the same in the long mode. It's stretches, squeaks and make them sound like some sort of soft floating glitches, which is not what we want. But if we lower both sensitivity and the reduction, we get more natural squeak reduction. Let's move to the last tab and also switch to the PEC selection. The peaks are tall and thin and it sounds very sharp. Make sure you listen to this piece loud enough. These are the plucks of the guitar pick hitting the strings. This noise can be created not only by the guitar pick, but also by the nails. For example, in classical or flamenco guitar music, or the fingerstick genre player uses his nails instead of the guitar pick, which creates the same noise. Now let's try to create a few previews and compare it to how it looks and how it sounds. The more reduction, the shorter these peaks, but also the longer the attack, the even shorter the peaks. Because of attack is fast, the peak is denser. And if the attack is slow, the pKa sort of scattered. It sounds a lot smoother and this is not the same if we just use a compressor or dynamic equalizer. This tool is lean to detect only specific pluck sounds, so it preserves the other information. 25. Interpolate: Interpolator is probably the simplest our next module, but it's a very important and useful repair tool. What does it do? A recreates the sound wave based on it's surrounded content. In other words, it's smoothers a crumpled piece of the sound wave which helps get read of the clicks and crackles. Let's go to the examples so you can see what I mean. Let's listen. I always thought something new would change my life. There is a very obvious click in the middle of the phrase. Here it is. But we don't need a spectrogram for this one. Let's go all the way to the left. Now if we zoom enough, we can see what's going on here. The sine is going pretty smoothly until here, and then it starts jumping until here. Let's zoom a bit more until we see the individual samples. And we don't have to before in this example, I'll count them. So I counted 27 samples in this Click. It can be a bit more or less depending on where we start and until where, but approximately it's 27 of them here. The interpolant module is able to synthesize or recreate up to 400 samples at a time, which is a pretty short piece relative to the whole audio, but pretty long range relative to the single click. The module has only one slider that adjusts how detailed we want the new sign to be compared to the damaged piece. Let's try the lowest and the highest value. The smallest value, it's just connected before and after the click with a pretty straight line. And in the highest value it tried to recreate the behavior of the sine wave. The click in this example is very small. I always thought something new would change my life. I always thought something new would change my life. So in both cases, the results sound perfect. But if you hear some artifacts after the processing, try to change the quality value, it won't take too much time since the processing here is instant. 26. Mouth De-Click: Mouth D Click is one of the most frequently used RX modules when working on podcasts, audio books, and other kinds of voice recordings perform being close to the microphone. These are specific type of noise created by lips and the tongue. We already covered the general D click module. There is a more specific tool learned to deal only with clicks and smacks coming out of the mouth. Let's compare these two modules. They look identical except this algorithm column, it's missing in the mouth to click. Why? Because the General de click module is designed to deal with clicks that have a different nature. And mouth D click already knows by default what kind of problems it has to search for. But there are three elements that are the same. Sensitivity where we adjust how deep it will search for the clicks. Frequency skew, where we indicate the frequency range priority, whether it's a low frequency range or the high frequency range. And the click widening, we regulate how much information we want RX to consider about the click. It helps when working on clicks with tails. The example we're going to work on is an ASMR whisper. I chose it because it contains a lot of very audible mouth clicks. Let's listen. They turn the music down, but it was still, it was quite. Now let's try to compare it with the default settings processing. They turn the music down. Still, It's quite honestly. This sounds much better, but we can still hear a lot of the clicks, though it's already a matter of preference. Let's try to maximize the sensitivity to clean the leftovers. Let's also drag our frequency skew slider more to the high frequency range because it's a female voice which is pretty high. Also, let's widen the click range a little. Let's compare. They turn the music up, but it was still it was still quite as loud. I don't hear mouth clicks anymore, but I can hear slight distortion in the places where the clicks were removed. Like here. For example, some sort of like freckles. In this case, it will be helpful to polish with the D crackle module. Let's render an open that a D crackled. Just like this with light settings. They turn the music down and up. It was still, it was still quite as well. The original. They turn the music down and up, but it was still, it was quite as after mouth **** Blick, they turn the music up, but it was still, it was still quite as loud. And after the final polish with the crackle, they turn the music down a bit, but it was still, it was quite as well. Awesome. 27. Music Rebalance: Music rebalance is one of the most popular RX modules of lung did not only sound engineers and audio editors, but also all kinds of musicians, singers, DJs, and so on. What's so special about it? The module allows us to separate a musical track into vocal, bass, drums, and other elements. We can change its volume in the mix. We can cut one or several instruments out of the mix or leave them only. Let's take a look at the module. Here are the controls of the four components. We can change the volume of the vocal, bass, drums, and all the other components that don't belong to any of these three categories and combined here. First, I want to say that the module has three quality modes. And if we choose the best one, we should get ready for the extremely slow processing. But it's definitely worth it. And I don't think I need to waste your time on reviewing these two modes because obviously the only benefit is faster processing. I'll be working on the best quality mode and just speed up the video every time my process something. For this episode, I have a lovely modern country song that includes all of the mentions components. And we're going to experiment with the chorus and touch the busiest part of the song. Let's listen. I think that's enough. We can already understand what this song is built off. And the second part of the course is pretty the same. Before we change the volume, we can solo any of these components and listen to them separately. I'll create a preview of each of them by clicking the Solo button and the Compare button so we can listen to these components one after the other. Dutton, the first goes the vocal. And we can see in the spectrogram it became much cleaner, more readable, and the waveform is also lessened. Let's listen. This is a very busy mix. A lot of instruments intersect. And so even though it's far from perfect, arcs did a great job by separating it. And I honestly don't know any tool today that can provide a better result. The basis is pretty doll, but it's also not in the first chair in the mix. I think we should be thankful for this result. It's audible and you can easily identify the notes and repeat them. If you're a bass player, for example, and you're using this tool to get a clue of what the base part is in the song. Next is a percussion separation, which sounds pretty impressive. The last we will hear as all the other components of the song which don't belong to the vocals based or drums. What we can hear is mostly clean electric guitar and some sort of barely noticeable pad. Next, what we can do here is less than the volume on one of these components or even minimize its presence. Let's dry vocal. That's probably the most attractive thing about music rebalance creating a karaoke that has never been this simple. And of course we can remove the other instruments available from the mix if we need to. We can also increase something, for example, base. But this track is mastered and it doesn't have much headroom, so I'll need to lower the other instruments to increase something without causing clipping. Now it's more basi. The separation slider regulates the accuracy of the separation. The lowest values we can still hear the other instruments. And in the highest, the separation is more spotless. We can also separate these four components into these new tabs and work on them apart. 28. Spectral De-noise: Spectral de-noise is one of the main and most effective aurochs tools and removing static or low dynamic noise. It also has the biggest module of this kid in terms of the number of controls and settings. But even though it looks complicated, you don't need a PhD in acoustic science to master it. Everything here is pretty simple. Besides, we're already familiar with some of these controls from previous modules. The major part of the module is taken by the spectrum display. The representation in this display is not like what we already have to get used to seeing in the main RX window. Here it's more of an equalizer visualization type, horizontally located frequencies and vertically loudness level. But what we have in common is with the main RX visualizer is Zoom and scale types. For example, I choose extended log and I get a lot of the low-end focus. I choose linear. And now all of the frequency ranges are shown equally and the same with the decibel scale. We can see the force spectrum graphs and one curve that we can manually change. So what are those? The first, the gray one is the spectrum of the raw signal, unprocessed signal, which is currently entering the spectral de-noise. The second, the white one is the spectrum of the output signal, process signal, which is currently exiting the spectral de-noise. The third, the orange one is the amount of noise that was detected while learning or wall time adaptive skinning. The fourth, the yellow one is the amount of noise that will remain after the processing, including all of our reduction settings. And there's also a blue curve that can be changed right into the spectrum display. What do we need a form? Let's come back to that in a minute. First, I want to explain the meaning of these two sliders. With the use of the threshold slider, we can set the level where it starts to reduce noises. It's not that sensitive, so it doesn't require much precision. The higher the threshold, the more noise we can detect. The reduction slider is how much of the detected noise will be removed from the signal. The higher the value, the harder the reduction. So back to this blue curve. When it's straight, all the noise from the frequencies in the learned profile will be reduced following the actual noise profile specifications. Changing this curve lets us specify the frequencies where we need more reduction and where we need less of it. For example, I pulled this curve down here. So here at the sub bass frequencies will be the hardest noise reduction. If I rise this range, let's say around two k, There'll be the smallest amount of noise reduction. We can create it up to 25 points and direct the curves between them in any way we want. We can smoother this curve if we hear the difference between the reduction regions is too abrupt. It is also possible to reset all of the changes on this curve or even turn it off if we don't use it. We can also turn it off here, just like all the other spectrums. Here we choose between the speed and processing. Here we can try to find the artifact balance. If we keep the artifact control at 0, the processed audio may sound a little like a low quality MP3. And if it's on maximum, it may sound a bit dull. This artifacts lighter is present in most of the RX noise reduction modules. You should probably already know what I mean from previous episodes. I assume you already have a clue from the context or from looking at this module that spectral de-noise has to noise detection Options. Learn, which means we have the select a piece of single noise without the main signal. So the module will remember it and reduce this noise from the main signal. And adaptive mode, which means the noise profile will be generating in real-time while previewing or rendering. When we work in adaptive mode, we can adjust the Learn time. This means that the module, we'll look ahead in the amount of milliseconds and accordingly either adjusting to the upcoming noise changes. But the more look ahead we set, the longer it will take the preview or render. The module also has a bunch of advanced settings. But before we go there, let's try the main setting we've just covered. I've prepared two examples for this episode. The first one is already familiar to us, buzzing electric guitar. We've already tried to remove this buzz with the use of the guitar denoise module, but we didn't get any convenient result back then. Let's first try to keep everything by default and learn this pause between guitar riffs. Let's listen. Yes, the bus got significantly quieted other, but it's still noticeable. Let's increase the reduction level. Cool, but there's still a little buzzing left. Let's minimize the artifact control, which often lessons the reduction. Let's compare it with the original wonderful result. Now let's try the adaptive mode on this piece. Let's switch to the second example and tested here. It's a very busy, rainy forest ambience here that is destructing a lot of the dialogue. Maybe our long time watchers have seen us do to walk stove and I'm always standing on a ladder and like salmon berry stumps sticking out. So like if I fall and I didn't get sphere. So this is, let's try the adaptive mode here a long time and watches have seen us do the walk stove and I'm always standing on a ladder and like salmon berry stumps sticking out. So like if I fall and I get spheres. So this is, let's keep the artifact control at a minimum and rise the little threshold and reduction sliders. Here a long time, watches have seen us do the walk stove and I'm always standing on a ladder and there's like salmon very stumps. It's like if I called and I get spheres. When the learned time is minimized, the previewing goes up almost instantly. But what it's maximized, it takes like three to four times longer to create a preview. Though the voice with the maximized learn time is much stronger. And clearly, if we compare it with the original, we can see that it's a great result. Long time watches have seen us do the walk stove and I'm always standing on a ladder and like salmon berry stumps sticking out. So like if I called and I get scared. So this is maybe a long time watchers have seen us do to walk stove and I'm always scanning on a ladder and there's like salmon berry stumps sticking out. So like if I called and I didn't get spheres. So this is there is no chance to remove all the ambience without a huge damage to the voice. But it's also what noise reduction and audio prayer in general is about. We can turn good audio into great audio. And if we have bad audio, we can make it sound good or acceptable. But we definitely can't make this rainy forest dialogue sound like a spotless studio recording. Let's also try to learn in this recording. Here's the pauses between phrases where we can learn. A long time. Watchers has seen us through the walk stove and I'm always standing on a ladder and it was like salmon berry stumps sticking out. So like my fault and I also did a great job. Now let's run through the advanced settings algorithm behavior. This is about artifact control. Roughly speaking, the more we increase the smoothing value, the more sensitive the main artifact control slider becomes. Here we choose the algorithm that RX uses to smooth the artifacts. Again, simply saying from the top to bottom, from the speed to quality. If we untick multi-resolution, we can change the Fast Fourier Transform size. The lesser the value, the fewer frequency bands will be used to noise reduction. It's less precise than having more beds. But in this case, the module will be adopting to the noise changes faster. And the opposite with having the high values here, we will get more precision, but a slower adaptation for the noise changes. And if we enable multi-resolution, the module will automatically choose the most appropriate FFT size for the Learn part. Noise floor, synthesis and Enchantments are pretty similar functions since this is responsible for generating high frequency harmonics that could be removed with noise. This helps to bring more life to the signal, making it less LDL. And the enchantment is a technology of finding important harmonics and enhancing them to stand out from the noise. Let's try to make a comparison with these two sliders in lowest and highest values. Longtime Watchers has seen us through the walk stove and I'm always standing on a ladder and just like salmon, very stumps thinking else like my fault and I guess gears here a long time. Watches have seen us through the rocket stove and I'm always standing on a ladder and like salmon very stumps thinking like my fault and I excuse. Even though it's not very audible, we can see this additional information appeared in hatched. This information above 15 k was synthesized. Masking is also a very smart function that adjust the amount of noise reduction by analyzing the signal. If there is a noise, but the below the level we can hear, the module won't be removing it to preserve useful information. The more this value, the more inaudible information will remain. Whitening is a process of smoothing the output noise curve into something that looks like a white profile noise. Take a look at this yellow scale, and let's compare and listen. Long time watchers have seen us do the walk stove and I'm always standing on a ladder and like salmon berry stumped sticking out. So like my fault and I get sphere. So this is a long time. Watches have seen us do the walk stove and I'm always standing on a ladder and there's like salmon very stumps sticking out. So like my fault and I get spheres. The definitely sound different and it's hard to say which is better. It all depends upon what we need to achieve. In 0 whitening we could hear more complex noise. While in maximum whitening, the noises more audible on a high frequencies only. The dynamics function are similar to what we have in the compressor. The NEA just saw sharp and abrupt will be the noise reduction. The lesser values provide a more soft and smooth reduction process, while the high values make reduction noise more abrupt. And the release time is how long the reduction will last after the detected noise. The higher this value, the slower the audio will return to the original loudness. 29. Spectral Recovery: We've already covered a lot of different causes that can lower the quality of our recordings. One of the most popular and growing and demand recording types is virtual call recording. Any kind of interviews and podcasts. But all of these programs like Zoom, Skype, Messenger and so on compress the signal to make it lighter. Because the less information we send over the Internet, the less likely we are to get glitches, delays, or even the loss of letters, words, and sentences. All these communication programs are pretty similar algorithms that include high frequency cuts. As the result, we can clearly understand the conversation, but the voices are often sound doll and lifeless. The spectral recovery module helps us to enhance the speech by generating missing contact from four kilohertz to 20 kilohertz. Before we begin, I want to note that this is very specific and sensitive tool. And in most cases we won't be able to achieve a noticeable enough result without creating noise on the high end. But sometimes this is a lifesaver. The module automatically leering the selection or when nothing is selected, it's learning the entire audio. We can see that our example is cutting around eight K. And this is where we can see this fall in this display frequencies are located horizontally and the game vertically. Both scales are zoomable and we can see that our frequencies start decreasing somewhere between 7.707.8 k. The module detected automatically. And that's why we see this gray range over here. This range spectral recovery will be generate new frequencies based on the incoming signals. Sometimes when we switch between selections, it may not update the settings automatically so we can press the Learn button to make sure our settings, the current selection sets only the cutoff and the smoothing parameters. These two remaining sliders are meant to be adjusted only manually. We will come back to these sliders in a second. After the Learn button, we have this spectral patching checkbox. If it's enabled, the module will fill up these significantly frequency holes at a Raul, all the frequency ranges even below for k. This information will be generated by sampling nearby areas. Let me demonstrate all cut, let's say at 1k render. And if we see that it generated something here as well. If we turn it off, the module generates only the information above the level set here. But don't mix the purpose of spectral patching with the general purpose of the module. When spectral patching is enabled, it scans only the nearest area and create similar information to fill the holes. But the general purpose of the module is to generate a logical continuation based on the whole frequency range that is below the cut. The first slider is the amount of information we want to generate. The cutoff is where we choose to range of this gray zone. This is the range where the signal will be generated from four K, 20 K. Here we choose the balance of the vowels and siblings in the generated signal. For example, if we deal with the result of heavy decreasing, we would probably want to try to increase the number of symbols and listen to how it sounds. The last slider smooths the difference between the original audio and generated information. Let's listen to the original. Mainly in Ethiopia, I was a country lead working directly with small and growing businesses in the dairy sector. Let's try to add some high frequencies to like, usually starting with the default settings. Mainly in Ethiopia I was at a country leads working directly with small and grow businesses in the dairy. Mainly in Ethiopia was a country lead working directly with small and growing businesses in the dairy sector. I don't feel a need to change the amount of generated information as well. The cutoff position. What I want is to try changing the balance between vowels and singlets, as well as the smoothing amount. Let's try to maximize vowels. Mainly in Ethiopia I was at a country leads working directly with small and growing businesses in the dairy sector. I liked the waitstaff floating at very high end of every S and D letter. Now let's try the other way around. Mainly in Ethiopia was a country leads working directly with small and growing businesses in the dairy sector? No, In my opinion, the previous one was better, swollen maximize the vowels again. Now let's try to change the smoothing percentage. I'll move it to the middle. Mainly in Ethiopia, I was a country lead working directly with small and growing businesses in the dairy sector. Now let's try to maximize a mainly in Ethiopia, I was a country lead working directly with small and growing businesses in the dairy sector. Now it's softened mainly in Ethiopia, I was a country lead working directly with small and growing businesses in the dairy sector, mainly in Ethiopia, I was a country lead working directly with small and growing businesses in the dairy sector. 30. Spectral Repair: Throughout the course, we talk a lot about the static background noises and covered a lot of different ways to deal with them. But we haven't talked much yet about what to do with the noticeable dynamic noises that may occur while recording. Let's listen to these couple of examples I've prepared for this episode. The first one is a dialogue in the center of a big city. This is Black Dragon. This is from Wales, a lot of cider. It's not the same as the site of inter-cluster front sight of his just fermented apple juice reading. We want to preserve the ambiance in general, but there is a loud car horn that stands out and distracts from the voice. Here's the first time at horns. This is black dragon. The second is short. And here's the third one. Minute cluster will try to remove it or at least minimize them. And in the second example is the dog barking while recording and acoustic guitar. Here's the first. Here's the second one. It's important to be able to detect these unwanted elements on the spectrogram because this is how the spectral repair works. We have to select the unwanted element to remove it. So let's switch back to the first example and open the spectral repair, which is designed specifically for this kind of sudden interruption. As we can see, the module consists of four tabs, which are the four different ways to deal with noise. Let's start from the first one which calls and attenuate according to the name. This tool is meant only to reduce the loudness of the selected noise to make it less noticeable. For it to sit inside the static ambiance and don't call him the attention. The first thing that we can set here as the number of bands, which means the number of frequency cuts the module will perform. The small number of bands fits better to the short sounds and the bigger number for the long ones. The bigger number of the bands provide more frequency precision and requires the wider surrounding region to learn from. Why does it need to learn? In this case, the attenuates section, the module scans the area around the selection to make the selected elements sound less loud and noticeable the way its surroundings are. Here we choose the length of the area that's spectral repair will analyze and learn from. And this is how strong we want the production to be. The multi-resolution mode allows for better frequency resolution for the interpolation of low-frequency contact and better time resolution for the interpolation of high frequency content. Using this slider, we specify words going to learn from the left side, from the right side. Or if the slider is in the middle, we will learn equally both from the left and right sides. We can see how the selection is moving from left to right, where we can change the position of the slider. This is the area that the spectral repair will analyze and learn farm. But we can learn not only from left to right, but also from the higher and lower frequencies of the current selection. For example, if we don't want to interpolate anything that is happening before or after the current selection, we can switch the mode to vertical here. And now it's going to analyze the frequencies instead of time. Now when we move this slider again, it's expanding selection nor left and right, but up and down. We can also combine vertical and horizontal interpolations. Now let's get into the practice. Let's listen to the first example. Once again. This is Black Dragon. This is from Wales, a lot of cider. It's not the same as the site of inter-cluster front sight of his just fermented apple juice reading. This horn has a pretty straight nature. I mean, it doesn't change its toner character. And this is exactly what we can see here. The bright lines are the harmonics of the car horn. So instead of selecting them one-by-one with the acyl or the brush. I'll use the Magic Wand Tool. I'll click on the first noticeable harmonic here, and then I click this icon and click on all. This means the RX will automatically detect all of the harmonics from my selection. The maximum it can detect automatically as ten. Here are all of them now selected. After we selected, It's important to listen to the selection only to make sure we haven't selected any of the main information. As we already know, there are two playback modes and RX, the regular playback, and the frequency selection playback, which is the one we need right now. Yes, we don't hear anything but the car horn. We can see that the harmonics actually appear a bit earlier. Here are the beginning, but it's pretty blurred, which is why RX didn't detect it automatically. What I want to do now is to take up brush and holding shift at these heads to the detected harmonics. This size is too big for the task. So I long press the brush icon to change its size. When we hold shift, we can see this small plus sign, which means we want to add the selection. But if we hold Option or Alt, we can see the minus sign, which means we want to exclude part of the selection. It's very handy when we've selected something by mistake. Let's turn on the multi-resolution, increase the number of bands and accordingly widened the surroundings and maximize the production strength since the horn is very loud. Let's listen. This is Black Dragon. This is Black Dragon. Fantastic. The horn is still there, but it's mixed up with the general ambiance. And if we weren't working on this audio, we'd not notice it already. Now let's switch to the Replace tab and check how this mode will handle the situation. How does it work? In the previous attenuate mode we can quiet and these harmonics and the replace mode, we can totally replace these harmonics with the surrounded content. Fortunately, I don't need to select the horn again. Just select here original audio and it brings all the selections back. Let's try now. Black Dragon. And let's maximize the number of bands. And why didn't the interpolation area? This is Black Dragon. This is from West. Well, yeah, it also does the job with the attenuation provides a more natural sounding result in this case, for the pattern tab, Let's switch to the second example. Here are the two places where the dark barks and let's try to do something about it. In the pattern mode, the module is searching for the piece similar to our selection and just duplicating a. And all the controls here are similar to the previous tab, except there was a waiting and now we have a search range. In search range we specify in seconds how far we want to search for a similar piece. Let's select the first bark and try to tweak the controls. To be honest, it's almost impossible to know sure what values were suited in this particular case. In my case, this is the best result I can get here. There is a small, barely noticeable glitch left, but it's a great result anyway. Now let's try to remove the second bark using the last tool. Partials and noise is the advanced version of the replace mode. Using this tool, our selection not only will be replaced with the content from the nearby area, but the new harmonics will be generated based on the nearby area. This is why in addition to the previous controls, we have a harmonic sensitivity slider here. It adjusts the amount of generated harmonics. We can have this test harmonic sensitivity chalk box that allows us to listen to the generated content only. For example, I'll leave everything by default here, tick this checkbox, and click Compare. And now we can listen to what we have instead of the dog barking. Let's try to apply these default settings here. No, let's lessen the harmonic sensitivity and increase surrounding region. Yes, that's much better. Let's compare it to the original. 31. Voice De-noise: Voice denoise is one of the most frequently used RX modules. He uses a very smart algorithm capable to detect voice or a musical content and separate it from the noise using 64 band-pass filters. In some ways it's similar to the spectral de-noise that we've covered a few episodes earlier. But unlike the spectral de-noise, the voice de-noise has a much smaller amount of controls, which makes it more attractive for beginners. But even though voice denoise has fewer parameters to set, sometimes it could provides us with the best results. In the center of the module, we can see the spectral display. The frequencies are located horizontally and the level is located vertically. The gray spectrum represents the signal that is entering the voice denoise, the white spectral, a signal that is exiting the voice denoise and the blue curve is a threshold level. The module has the Learn and the adaptive loads. To get any results from using the Learn button, we need to show the voice denoise the pure noise without the main information. For example, here I have almost 1 second of the ambiance before the voices coming out. I select it, press learn, and we see how the threshold line has been changed. This line has six adjustable points strictly bound to certain frequencies. We can move them up and down, changing their levels, but not the frequencies, which means they are locked from moving horizontally. The higher the dot, the higher the threshold on this particularly frequency range, which means we will have more noise suppression here. Let's learn the selection, maximize the threshold and the reduction level and try how does this work? I move this dot down, releases the threshold here for good, Relax to move this, no hustle on the streets, nothing. It's just a good Relax tool. And it releases the frequency range at around 600 hertz and so on. This is the general threshold level and these dots are the adjustments to it. But if we switch to the adaptive mode, this curves become unavailable to us because now the voice denoise is automatically adjusting for it. Any changes in the audio. Here we specify the type of content we're working on. In my case, it's dialogue, so I'll leave it as is. And we can also choose a filter type. The surgical type provides more reduction, but sometimes when there is a lot of short pauses in dialogue with the loud background, we can get sort of a broad loudness changes and the gentle mode provides lesser noise reduction in exchange for a more smooth reaction to the changes. Now that we already know how everything works here, Let's try to lessen the background noise using both the manual and automatic modes. Let's keep these two sliders on maximum because the noise here is extremely loud. One more advantage of the voice denoise is that it's pretty fast so we can hear the result in real time. And this is exactly what we need to hear the changes in real-time. So once we are happy with what we hear, we will press the Compare button. Let's go. No hustle on the streets, nothing. It's just a good Relax tool. No hustle on the streets, nothing. It's just a good relaxed, chilled like that's a great noise removal, but the voice now feels a bit distant. What we can do here is to play with these three middle dots by releasing them a little. Since they're there, What's whatever the voice is located? No hustle on the streets, nothing. It's just a good Relax tool like hustle on the streets. Nothing. It's just a good relaxed, chilled. No hassle on the streets, nothing. It's just a good Relax tool. Just like this. There is a bit more noise now, but they'll voice is much more noticeable already. I like the gentle mode. In this case, I click compared to save these change settings. And now let's try the adaptive mode. No hassle on the streets, nothing. It's just a good Relax tool. Tap. No hassle on the streets? Nothing. It's just a good Relax tool. No hassle on the streets, nothing. It's just a good Relax, true by both surgical and gentle modes provide excellent voice clearness and we may want to release the threshold a little for the ambient is not to be chopped that much and not to sound abrupt. 32. Wow & Flutter: In this episode, we're going to fix pitch variation similar to what we can hear on very old venule and broken tape players or an emulation in most of the lo-fi tracks. Listen to them to the example for a better understanding of what I do to mean. So we can hear this sort of Vibrio at the end of every chord. And we can easily see these waves on the spectrogram. In this case, the wow effect is pretty fast, but it can be much slower and we would see longer waves here. Let's take a look at the whirlwind flutter module. We have two types here. And the only difference between the while and the flutter is that the law is a pitch variation with frequencies up to five hertz and flutter up to 40 hertz. This means that we can use the wild tab first, lower pitch variations and the flutter tab for faster ones. However, we can specify the temple of the wow effect with the use of these three modes. Fast, medium and slow. Fast suits for the variation with frequencies from two to five hertz, medium from 0.5 to two hertz. And the slow mode is meant for variation slower than 0.5 hertz and they aren't available in the flutter tab. This is how deep it will scan the audio for the pitch variations. We can also correct the global pitch offset when the whole piece of the audio is out of tune. For example, we can specify the oriental frequency here and the audio will be turned to it. If we take this checkbox, no correction will be applied with the wow or flutter will be highlighted for better visualization and auditability. Just like this. Okay, So if we aren't sure what mode to choose the quickest way is to try them all. What you can see here that the waves are pretty frequent. Why choose the fast mode? I click preview and we see that the difference already. Let's listen. Perfect. 33. Azimuth: And now we're moving to the utility section of the plugin panel. The first one here is the azimuth tool. So let's figure out what it's meant for. Let's listen to the audio. We can hear the left channel is louder and earlier than the right one. And let's click this icon to view the channel separately and move this slider to the very left to see the pure wave form. This is how different the channels are. The left channel starts much earlier and its amplitude is much bigger. Of course, we could just separate this stereo into to monitor channels and then synchronize their timing and balance the volume. But it would take much more time. The fastest way to fix it is to use the azimuth module. Let's open it. Here we set the volume of the left and right channels. And here the delay relative to the current starting point. If you don't know where to start, you can try to use the suggest button, but it's not always the correct. Let's try. Here's how big the things to the volume difference between the left and right channel, it's adjusted, increasing the volume of the right channel by 7.7 decibels, which makes a lot of sense. But here, somehow it's adjusts, delaying the right channel even more on 9.6 milliseconds. I already know that it's not going to help, but let's listen anyway. Yes, so both channels have become equally loud, but the delay is still here. It's still here and it's huge. So what we need to do is to pull the right channel to the left and the left channel to the right for them to meet in-between. Let's reach the maximum and listen. Yes, this sounds much better and sounds pretty balanced. But we can still see a bit of delay. It's not that noticeable, but if we want to synchronize the channels even more and we can render it twice. On the second render, we don't need the volume adjustment anymore since it's already perfectly balanced. Here we go. Of course it sounds more like a mono now. But this episode is not about mixing and mastering. It's about how fast and easy to Piazza moon is solving the problem. 34. Dither: Othering is the process of adding noise to the audio signal one Loring, it's a bit daft. It helps to avoid sinewave errors and to preserve the dynamics of the audio while converting bit value. The most common bit depths value today are 322416 bits per sample. However, for the well audible difference in this episode, we will cover 24-bit audio into 8-bit audio. I prepared a very simple tune originally created in 24 bits. This is how it sounds when converted into 8-bit without dithering. Let's open to either module. Our x use it isotopes m-bit plus noise sharpening algorithm, which is based on the psychoacoustic essentials. Its main point is to generate more noise on the less audible frequencies and less noise on the frequencies that human ear is sensitive to. We can see it here on the spectrum display. This yellow line represents the amount and the shape of the dither. Let's select eight here. This will be the bit depth. We're going to convert our audio two. And here we choose the shaping type. Take a look at how the spectrum changes when I switch the shapes from none to ultra. It's a little high at the low end, all the mid-range and high mid-range is very quiet. And then a lot of noise starts from 14 k, where a lot of the audio devices won't even be reached. And this is the dither amount itself, the highest value, the noise we can get from. But since the noise here is meant to improve the quality of the conversion, It's not always a good idea to keep it low or even switch attend none. Here we choose when dithering will stop. Never, which means the static noise will go through the entire audio from the beginning to the end, no matter what. During silence, which means the dithering will follow the signal and stop at every pause this signal has. And when quanta sized, the dithering will be suspended in case the other dithering is detected. We may want to enable this checkbox when using high noise shaping modes. This will preserve the signal from clipping. And the last option here becomes available only on the load dithering amount and the noise shaping settings. And lets us enable additional harmonics suppression when there is not enough dithering to guarantee the conversion without distortion. Now let's try to check how does 8-bit sound when dithering. This piece is very short and simple, so we can easily here the amount of white noise. Here's our starting point without any dither and noise shaping. Now if we choose the ultra noise shaping and low to their amount, it'll sound like this. In this case, the load dither amount works perfectly. And of course we hear some HIS, but this is the best results converting the audio down to eight bit. Of course, most of the time we decrease it to 3224 or 24 to 16 bit. And then noise there is barely audible. 35. EQ: In this episode, we're going to cover the RX equalizer before we dive in. All note that this is just a regular digital equalizer without any unique features that can make you've confused. If you already have experience using any other EQs, nothing will surprise you here. With the use of the equalizer, we get control over the specific frequencies and frequency ranges. We can make them louder, quieter, or completely remove them from selected audio. The equalizer works the same way on any audio. What do we have here? The middle part of the module is the frequency display. But unlike in many other EQs, we can't visually monitor anything here. We can set our frequency bands and filters here, but we can't see the frequency spectrum of the audio. Here's the zoomable frequency scale. Here is also the zoomable decibel scale. How does it work? If this 0 decibel line is absolutely straight, it means that there are no changes yet and the audio will say the same. To be able to change there any frequency, you need to enable at least one of these tools. We have here six bands and two filters. There's high-pass and low-pass. The first one, high-pass. If we enable it, we can remove frequencies from the low to the high. Here we choose how sharp or smooth we want the shape of this filter. The gray zone is where the frequencies were removed. If we turn it off, the sound will come back to the original state with the settings of this tools are saved. And if we turn it back on, it'll automatically returned to the last position. If it was a high pass filter in some equalizers, it is also cut a low cut filter because it cuts the low frequencies and passes high frequencies. The absolute opposite is true too. This as a low-pass filter, it does everything the same but in opposite direction. We can cut the frequencies from right to left, from high to low. We also have six bands that allows us both the cut or boost frequencies anywhere from 20 to 20 thousand hertz. For example, I want to cut here, boost here, and then cut here. We can control the amount of boost or cut by dragging them up or down. And we can also change their width by dragging this dash left or right. All these things we can also do here by clicking on these numbers. Here's the frequency, here's the game. And here's the Q, which is the width of the frequency band. This band is also called the bell, but we can change it to the lower high shelf. We use it when we want to increase or decrease the gain of the lower high frequency ranges, we don't want to cut them completely as we do with the high-pass or low-pass filters. I'll switch back to Bell. We can also change the sharpness of the bands and filters. The less the value here, the sharper color boost, and the higher the value, the smoother it is. This equalizer has two modes, digital and analog. If we switch to the analog mode, the frequency precision becomes inactive because this mode is an emulation of the analog physical equalizers, which are not that flexible as the digital ones. Here are the shapes start from wide to narrow, and this is something we cannot do in the digital mode. So both modes are useful the same. And the last thing here is a handy monitoring feature. If while previewing we hold Option or Alt and also click and hold a live band. We can hear the selected frequency range. And after we do so, our x remember is the width of this bad. And now we can also hold Option or Alt. Click anywhere we like and hold are moving. Listen to any frequency ranges are any amount of decibels. The next handy feature is to hold Shift while changing the game, locks the horizontal movement so we can't accidentally change the frequency while moving it up or down. This really helps a lot when we look for resonating frequencies with a high for precise thin bed. Once we've found it, It's very important to cut it without moving left or right. 36. EQ Match: With the use of EQ match, we can easily learn the equalizer settings of the one audio and apply them with another. The module is incredibly simple. We just said how precise we want to analyze the frequency spectrum of the selection and press learn. And here we see the EQ spectrum. Now we switch to the second tab. Press compare and see how it changed. This also works great with the single sounds. For example, with the drunk kicks sample. Here's the reference kick that we want to learn from. This is how it looks like. Here's the kick we want to apply to their settings too. That's it. 37. Fade & Gain: We already mentioned the fate mode earlier in this course. It helps to create a smooth beginning and the ending of the sound. When we were talking about decreasing the amount of breathing in the recording, I mentioned that there are few tools to do it manually and fade is one of them. Let's come back to that vocal speed. Let's try to make these breaths a bit quieter and shorter. Here's every new line starts with a breath. I'll repeat what I said the first time when we were listening to these vocals. The breaths are not the problem. It's more about the personal preferences of a producer and sound engineer. There is a million songs out there with heavier breaths, but we're going to reduce them for educational purposes. So the fade module has two modes. Baden, which smooths at the beginning and fade-out, which smooths the ending. In each of these modes have for phage shapes, our breaths or at the beginning of the phrases. If we choose fade in and the longer our selection is, the smoother the fade will be. It starts fading here from the beginning of the selection and until here, the end of the selection. So if I keep it like this, it will also fade the vocal, which is not what we need. Also looked a little bit of the breath. And now let's compare. Great. And if we select log, the fade starts much slower and much faster before the end. Since our breath is in the first part of the selection, it will be quieter than if it wasn't a linear shape. And everything is absolutely the same with the fade-out. Let's imagine this is the ending of the song and we need to fade it out. We select somewhere from here and until the end. Because if I end the selection right at the end of the phrase, it won't be that smooth if I expand the selection until here. The next thing we can use here at the game module, It's as simple as possible. We just select the range we want to change the volume on. And we specify how many decibels we want to increase or decrease on the selection. For example, I select this breath and let's say I wanted to decrease it on six decibels. I choose here negative six, and press Render. Done. I like using the game module to clean the pauses between the signal, whether it's a musical fall or conversation. Because if I just delete the pause, it will also remove this space between the phrases and the change of timing of this file. But if I use the game with hard reduction, let's say maximum, the timing will remain the same with the space won't contain any information anymore. For example, here's the mouth click. And I just cleaned it up like this. Very fast in handy. 38. Leveler: With the use of the leveler, we can quickly finalize the audio in terms of loudness and some of the additional features like S reduction, breath control, etc. However, this is not a Mastering plug-in. You won't be able to adjust the width of your mics, set the multi-band compression or anything like that. For this purpose, this isotope has a fantastic product name, ozone, which covers all of the aspects of top-notch mastering. The first thing that catches our eye when we open the module is the root mean square statistics of the current audio. I recommend you learn it to the RMS if you don't know what it's about because you'll face a lot of working in the audio post-production area. The total RMS, which is the root mean square, is the average value of the entire audio file. This is the highest value in the 50 millisecond range. And this is the lowest value also in the 50 millisecond range. Roughly speaking, this is the loudest value on our audio in the very short range. This is the quietest, and this is the average loudness. Everything in decibels. The leveler is a smart tool and it sets the clip gain to fulfill our settings. For us to see the clip gain, we need to click the View clip gain icon or simply press Command plus G or Control plus G. Next, we need to specify whether our audio is a dialogue or music, the level or uses different algorithms for both types. Next is the main slider, which we specify the average root-mean-squared level we want to reach. But to avoid clipping to provide a high compression transparency level, the output RMS level may want to vary from what we said here, especially on the high gain values. Responsiveness is where we set the reaction speed of the clip gain to the amplitude changes the lower the value, the sharper the clip game. It helps the shape the attack, but sometimes can make it a bit abrupt and even cause some unwanted artifacts. The higher values we shape the level amount of a single transient, but the phrases and sentences, with the use of the preserved dynamics lighter. We adjust the difference between the quiet and the loud sounds. The lower values provide us with the more original dynamics. The difference between peaks and dips will be more like the audio original. And the higher values provide maximum equality, which means the loudest moments will be quieter and the quietest louder. And the last two sliders as reduction and breath control or the simplified functions of the modules we've already covered earlier. We can adjust the level of the sharp syllabus and breaths or we can turn these functions off. We want to apply neither of them. Let's try it out. Here's the original audio. Where's my stuff? We didn't know that. My dad replied and winks at my mom. We can hear the last part a lot quieter than the first one. Let's say I want to reach around negative 16 decibels. Now it's negative 21. And let's say I want both parts sounding equally loud. So I keep the preserved dynamic slider minimized. But let's compare it with the slider on maximum. We can see that when preserved dynamics is minimized, both parts of the audio are pretty equal. And when it's maximized, the original dynamic is preserved and not much changed from the original. But in both cases the audio became louder in general. And if we render it now, we can see that these numbers have changed. Though we're reaching for negative 16. It's actually louder. It's also around negative 14 mill. In this case, we need to go lower to reach negative 16. Let's try negative 19 decibels. Yes, that's it. 39. Loudness control: The loudness control is also meant to change the volume of the audio. But unlike the leveler, it doesn't adjust the clip gain relative to the amplitude changes, it increases or decreases a signal equally. The loudness control is ELC, AFS focused, ALK EFS is the same as L UFS and the difference is only in the name elk EFS is loudness K weighted full-scale, and L UFS is loudness unit full-scale. This is the smart loudness calculation algorithm based not only on technical specific patients, but also on human loudness perception. The module consists of three parts. Here we set the target, the parameters we want our audio to meet. Here we see the current loudness specifications of the audio. And if we click this arrow, we opened the display with the parameters are visualized on the time and loudness scales. When we set a true peak, which is the highest amplitude point, we should remember about the digital analog conversion. As long as the audio stays inside the digital device are D-Day, I'll be you cell phone, streaming platform and so on. These peaks will remain on exactly the same point. But once we open and try to listen, the signal converts from digital to analog, which may have a slightly different amplitude. And if we keep the true peak at 0 and analog amplitude turns to be a little higher, we will get clipping. So for example, if you're planning to distribute your audio on YouTube or Spotify, keep your true peak at minus one decibel and you will meet their disabled TP requirements. The next target is the integrated elk AFS. This is the average loudness value. And as I said, it's based not only on decibel calculations, but is also considering the frequency levels since our ears have different sensitivity to the different frequencies. For this episode, I also attached a PDF file where you can find the true peak and the integrated alkane BFS or DFS requirements of the most popular streaming platforms. The next parameter is optional. We can turn it on and off and select the measurement types, short-term elk EFS, or momentary elk AFS. When we select the short-term elk AFS, we can set the maximum alkane EFS calculated in a 3 second time range. And if the momentary ALK EFS is selected, we can set the maximum FCFS calculated in a 400 millisecond time range, which is 0.4 of a second. Here we set in loudness units the amount of the discrepancies we allow discrepancies from the parameters we set above. The program loudness gate is enabled by default because it meets most of the common loudness requirements. What does it do when it's enabled? The low-level monetary signal is excluded from the calculations. In the second row, we see the specification of the current audio. This display is the visualization of these measurements integrated, short-term, and momentary. Let's say we're preparing the audio for the YouTube video. We set negative one and true peak or shorten DB TP. Then we set it to negative 14 integrated TFS, which is also one of the YouTube requirements. Let's say I'm not too sure what value I should use for the short-term or the momentary period. I simply turn it off for it to be set automatically. And we'll also keep the tolerance by default. Since we're dealing with the common loudness foundered, we also leave this function enabled. If we click this gear icon, we can turn on the high accuracy mode, which provides us with a higher processing. This will also make the processing a bit slower, but it's not critical for the short audio. Here we can export a log of the current loudness. Let's export it and open it here. Here are the timestamps on every 2.5th, the elk AFS values of the integrated short-term and momentary measurements. Okay, back to the module. We've already set our values here. So let's render. Here we go. The true peak now is negative one integrated LCA EFS is a negative 14. And these values are great because they're disabled here, but they are correct. And of course, these peaks of the waveform market now, because they were too loud. Although if we zoom in, we see that it's not clipping. The sinewave is rounded and there will be no distortion if we play it. But if we wanted to look more equal, we'd want to use the level of before. We discussed that in the previous video, it changes the clip gain to make the loudness more equal. Then we're coming back here and polishing the audio with the precise true peak and integrated LPF settings. 40. Mixing: With the use of the mixing module, we can change the precedence amount of the right and left channels in the stereo file. It has a pretty simple interface. Here we choose the presence of the left and right channels in the left output and here the same for the right output. It may be a bit confusing at first, but for you to understand, Let's play the audio. The piano and the Mallat are located at the very left. Whilst the plucking clean guitar and the pad or at the very right. Now these default current settings won't change anything. If I play, it's all the same. Because now it repeats the setting of this audio. It's 100% of the left channel and 0% of the right channel and the left output. And it's also 100% of the right channel and 0% of the left channel and the right output. If I move, let's say left channel in the left output from 100% to 0%, it will become inaudible, mute. The same with the right channel. Now let's maximize the right channel and the left output and the left channel and the right output. What's happening? The channels are now switched. Piano and malate are now playing in the right ear and then the plucking clean guitar and the pattern now playing in the left ear. But what happens when we go below 0? We also increase the presence of the channel, but only it's inverted now. So not only did we switch the left and the right channel, that we also flip them upside down. Here's the original audio. Here's when we just switched the right and the left channels. Here were the switch channels are also inverted. A very interesting and very useful tool when working with the stereo. 41. Normalize: The normalized tool is simply increasing or reducing the level of the audio until the highest peak of it reaches the desired value. For example, here's the loudest part of the audio, somewhere at 0 decibels fs. And let's say I want to normalize it to negative six decibels fs. I said here negative six, and here it is. Now the whole audio becomes quieter and as much where necessary for the loudest moments to reach negative six decibels IFS. It's similar to the gain module because both of them just make the entire audio quieter or louder. But the difference is that in the gain tool we set the amount of decibels we want to increase or decrease. Here we set the decibel fs value. It has three CH. 42. Phase: The face tool is meant to fix some of the waveform issues by rotating it left or right at up to 180 degrees. Let's open the module to see what's in it. Here are the main controls where we choose the number of degrees we want to rotate our left and right channels. If minus r0 rotated left if plus right. These values for the left and right channels are linked by default, they move together, but we can unlink them by clicking this icon, and now they're independent. We also have a suggest button that lets the module analyze the sinewave and suggest these values. We can also enabled the adaptive phase rotation, the face tool or rotate the waveform depending on how it changes throughout time. Let's take a look at this waveform, which looks a bit unusual compared to what we're used to seeing. But if we zoom in a little, we see this beautiful equals sine wave. This is how pure for a 140 Hertz looks like. We can also see that in the left and the right channels here are almost an anti phase. When the left channel sign goes up, the right channel sign goes down and the other way around. If it was perfect anti phase, the channels would cancel each other in mono and there would be no sound at all. But we can see that they are not perfectly against each other, which is why instead of the cancellation, we just get a quieter in mono. But anyway, we don't want this to happen. So we're changing the phase between the left and the right channels. Here it is. When the left goes up, the rate goes up as well, and we're not afraid of the mono anymore. Another use of this tool is avoiding clipping. Let me use the Gain module to create us small clipping somewhere here. Let's say four decibels. If we play, we see this box get red because now the clipping occurs. What do we do? We select this clipping area and rotate it on example negative 80 degrees. What happens here is that the high amplitude got divided and it doesn't reaches 0 decibels anymore. No more clipping. And what's important, it doesn't change to the way it sounds. 43. Plug-in: Now it's time to talk about one of our x's most interesting and powerful features, which is the module called plug-in. Let us use our favorite audio unit or a VST plugins rights inside the RX. Unfortunately, it doesn't support VST three yet, but I'm sure it will soon. So how does it work? We open the plug-in module. We click Select plugin, and I already have some of the plug-ins here. But if you didn't specify the plugin path yet, you won't have anything here. So the first thing you need to do is go here, Manage Plugins. We click Add and add the folders that contain our VST and AU plug-ins. Then we can click rescan in case the scanning didn't automatically start. We can enable this function if we want to group the plug-ins that we have at least one similar word in their names. Now after a while you will have seen here all of the supported effect plugins ready to use from the folders you specified. For example, I'll take this free soft tubes iteration now plug-in. It's now contained in the RX module shell, which is why we still have all these functions. It's working just the same as in our DAWs. This OTT plugin. I can preview it, make comparisons out of it, and of course, render when I liked the result. 44. Resample: The sample rate conversion or the resample tool is designed to provide flexible control over the process of increasing or decreasing the sample rate of the audio file. I recommend you dive into details about the sample rate as well as bit depth, what we were talking about in the dither episode. I'll try to quickly explain the basics. If we look at this sign, we see these thoughts and the spaces between them are individual samples. The more samples, the shorter they are. And accordingly, the more precise the sine wave, the sample rate is measured one per second. Which means if we're dealing with, let's say 44.1541 thousand samples are describing the sine wave in 1 second period per time. The sample rate determines the maximum audio frequency that it can be reproduced. Theoretically the maximum frequency that can be represented as half the sample rate known as the Nyquist frequency. So if we want to deliver the whole audible frequency range, which is 20 hertz to 20 kilohertz. The audio should be described at least 40 thousand samples per second. But practically there has to be a headroom for the transition band, which is why instead of 40 kilohertz, we have 44.1 kilohertz. Although this topic is much deeper than my explanation and there's some reasons to use higher sample rates than 44.1 kilohertz. The common practice is to record in higher sample rates. Let's save 48 or 96 K or even higher and then reduce it. I'm mixing, mastering or a distribution stages. This is when the correct resampling becomes very important, especially when it comes to down-sampling, which means decreasing the sample rate. Why especially downsampling, because when we decrease the sample rate, the upper frequencies that are now beyond the range needs to go somewhere. And they are shifting into the current frequency range, which causes the aliasing. Aliasing is defined as the miss identification of a signal frequency which can introduce distortion or other artifacts into the recording. Our current sine is 48 kilohertz, and here it's also 48 thousand is chosen, which is why it's written. No re-sampling, select a resampling rate. But let's say we need to downsample it to 44.1 kilohertz. I choose 44,100. And here it's suggesting the resembling options. The white line here calls the ideal filter. This is the edge with the new frequency range. And all of the information above this line will be shifted to the frequency range causing the aliasing, which is why we need to apply the cutoff filter. The cutoff filter is this yellow line and we can move it over frequencies by dragging the slider. Now we can see if we extend the cutoffs zone beyond this white line, beyond the ideal filter, this red line shows up. This is the aliasing that we don't want to have in our audio. That's why when we chose 44,100 here the first time, it automatically showed us the most balanced option. We have a very small frequency reduction here in exchange for the very few aliasing. But we can operate the sharpness of the cutoff filter with the use of the filter steepness slider. Now the reasonable question, why don't we just maximize the steepness and move it to the ideal filter? Because doing so will increase our pre ringing, which is also one of the artifacts that can be described as the reversed echo. And if for any reason we're getting this kind of artifact, we can try to remove it by changing the position of the slider. It helps us find the balance between pre-reading closer to 0 and post drinking closer to one. Also, when we were working with the recordings with the loudness close to 0, It's recommended to keep the post delimiter enabled to avoid clipping. D interesting function here is to change take Omi. If we enable it, no resampling will be made. But in the information section of the file will be written in new sample rate. It helps to prevent some of the length and playback speed differences when operating the file in different apps on different platforms. So let's try now to decrease the sample rate and let's see how it looks like. Let's go for the lowest value here, which is 11,025 samples. A second. Of course, there is no point to listen to this extremely short piece, but let's take a look at how many samples are year from now. From this to this. It's obviously not that precise as to who was around 48 K. Put the most important thing here is that the 11,025 sample rate can carry only up to around 5.5 kilohertz. And RX automatically adjusted its spectrogram for this frequency range and we can not go above it. 45. Signal Generator: The signal generator can be used to fix DC offset to illustrate different types of sine waves or to create some of the basic sound design elements. I already mentioned that test tone function while covering the preferences menu, we can hear the tone and real-time there. Here it works a bit differently. And to use this module, we need to load some audio or create an empty project. Let's create a new project. Let's start from the tones tab. We chose a waveform, frequency of the tone and how loud you want it to be. That's it for now. Let's render. We got here a typical 440 Hertz sine. Note that when our tab is not empty, when we have some audio here, we can now choose the pasting mode. When the tab is empty only the insert mode is available. But when we have some content here, we can choose how do we want this tone to be rendered in this tab? We can replace it, which means every new render will replace the current audio. Or we can mix, which means every new render will add new information to the current mode. For example, I choose Replace and choose here square. Now it's replaced. But now I change the mode to the mix. And now I choose here triangle. And let's say 700 hertz. Here we go. I'll undo it. What's interesting here is these additional functions. For example, I can make the slide from one frequency to another. Let's say I want a slide from a 100 hertz to 1000 hertz. I choose them here. And I can also specify the shape of the slide. Let's see how do each of them look like linearly log scale. In Mel scale. Another cool feature here is the modulation. Here we choose the amount of pitch jump, how deep the vibrator is going to be. Just take a look and listen, 10%, 30, 70% percent. Here we choose the speed of the vibration. It's five hertz by default, but let's set it to 20 hertz. The higher the frequency, the faster the vibrator. And let's compare it with one hurt. Great. We can also change the slide from increasing to decreasing. Let's say from five K to 100 hertz. Here's everything the same but only about the volume. This is the starting loudness and this will be the final. We can make the loudness increase or decrease with these sort of steps, just like we did the frequencies. And we can play with the anti-aliasing slider if we hear some of the artifacts. Now when we have some audio here, we can go to the silence tab and check how this DC offset slider works. Roughly speaking, DC is the zero-point of the waveform, and it's marked in our x with the infinity icon. This is important for the center of the waveform to lay right on this line. But sometimes it's off due to a bunch of the analog digital errors, and this is where we can fix it. The last tab here is the noise generator, just the same as the tone generator, but instead of one precise frequency here we get the noise. We choose the noise type and we see its frequency nature. The white noise is the same loud on all the frequencies, and pink and brown are more loud at the low frequencies. 46. Time & Pitch: With the use of the time in pitch tool, we can stretch the audio file, changing its temple, and also raise or laureates tone. Let's take a look. The first thing here is the stretch options. If we know the tempo of our file, we set it here. By default, it's 120 beats per minute, but I changed it to 112 BPM, which is the tempo of this audio. If we don't know our temple, the result will be approximate. But in this example we know our temple and now we can correctly speed it up or slow it down. For example, I want my new temple to be out 130 PPM. I set it here. And now we see that the Stretch ratio changed from 100% to 86 In 2% because this percentage is the coefficient between the initial temple and the new tempo. We press compare and see that it changed from around 11 seconds, approximately nine seconds. And we can hear it became faster. We can also achieve the same result if we just drag this slider. Let me return everything by default. Let's say I have no idea what my tempo is and I want to make the audio twice faster. I move this slider to 50% or type here 50. And that's it. These values now aren't correct relative to the current tempo of the audio, but they become correct relative to each other. The next use of this module is pitch shifting, which means moving the tonal elements of the signal up or down. I choose for this episode of pretty slow single vocal so we can clearly seats it's notes on the spectrogram. Here is the fundamental harmonic and all of the other right harmonics that multiply from it. We can listen to them separately by using the Magic Wand to select this button, to listen to the selection only. Before we start processing, we need to choose an algorithm for a. We have here high-quality solo and fast radius algorithms, which are the isotopes branded pitching technology. The first and the last are the ultimate algorithms for any kind of sound. And the only difference here is the priority is given to the speed and here to the quality. And the solo algorithm is meant for any kind of single toll. For example, if we're dealing with the violin or the vocal, which is in our case. Let's say we want to raise the voice on two semitones. We put here two, and that's it. We see the harmonics got higher and it sounds accordingly. We also have the advanced settings here. When the solo instrument is chosen, all of these functions are frozen and we have access to the adaptive Windows lighter. Simply put this slider, we can avoid some of the artifacts. For example, if we maximize it, we will get a sort of choppy sound. Let's try. The minimum position works great in this case. But if you're working with a fast solo part and pitching a lot, it may need a bigger window size in this case, to avoid the artifacts, if we switch to the radius, the window size will be set automatically and instead we'll have a bunch of other options available. Let's quickly run through them. Transient sensitivity, how deep the module was searched for the trends ins in this audio. It's better to keep this value low when working with this signal with highly audible transient to prevent them from Soundbeam to sharp. And the other way around when the transients are not clear enough, we might want to raise this value. Noise generation generates the noise instead of stretching the exiting one, It's most noticeable in some noisy or airy content types like a tale of a snare drum or any kind of siblings. In some cases, it may preserve the processed audio from sounding flanged. Synthetic pitch coherence helps preserve the original Tambora in the pitch solo elements. Although higher values can make it sound harsher, phase coherence can preserve the relations between the left and right channels. If we hear any changes in our stereo image, we can try to increase the value of the slider. And if we decrease it when working on multichannel with completely different instruments. The last section of the module is dedicated to format shifting formats or the resonating frequencies of the vocal tract. Roughly speaking, because of the formats we can tell, for example, the voice of the kid, from the voice of the adult, even if they have the same pitch. So let's try. We will increase four semitones, which are two tones to the voice without format shifting. And then with the formats also shifted to four semitones. Let's compare. Yes, In the first example, the voice got higher. But we still hear that this is the voice of the adult woman. And in the second example, it sounds like yo sunblock a kid. We can visually compare that when we shifted the formats. This information here also raised. Here, we adjust the strength of the format shifting. But to be honest, I've never heard of difference between the positions of the slider, so I keep it maximized. Here we adjust how wide will be the format detection range. Here we can also make two comparisons and make sure that nothing was changed. But I'm sure there is no useless functions in R x, so I'll probably didn't have a suitable material yet to hear the difference. But in general, format shifting is a fantastic feature and it works great not only with the voice but any kind of audio. And sometimes when experimenting we can get really cool unusual results. 47. Variable Pitch: Variable pitch is similar to the dialogue contour module we covered earlier. They have an almost identical interface and can actually perform pretty much the same tasks, but despite that, they have slightly different purposes, which is why a few of the controls are different. In my opinion, dialogue contour, it is meant to vary it a page in the original dialogue to change the expression method and the variable pitch is designed to fix some pitches shoes which often happen and are more critical in musical content. Although this preserved time function turns to the module into a great sound design tool as well. We will come back to that in a moment. So the workflow here is the same as we had in dialogue on tour. We have a straight pitch line that is located at 0. By default, we can create up to 25 points on it and raise them up or down accordingly, changing the pitch of the audio and the amount of semitones we pitched his written beside every point. We can smoother the curve between points or completely reset it. When the preserved time is disabled, the pitching is affecting the temple. The higher the pitch, the faster the playback and the other way around. Let's listen. If we enable this function, the pitch won't be affecting the same time in any way. These two sliders are already familiar to us from the previous episode. Pitch coherence helps preserve the original Tom Bray of the pitch solo elements, although higher values can make it sound harsher, transient sensitivity is how deep the module was searched for the transients in this audio, It's better to keep this value low when working with this signal with highly auditable transients to prevent them from sounding too sharp. And the other way around, when the transits are not clear enough, we might want to raise this value. Let's try now to fix this floating pitch. We have two issues in this audio. Here. The pitch bends down. Here it goes up. Let's select them and press M to mark them as the region so we can easily find them. It's not that deep, so I'll go with one semitone and the deepest part of it's somewhere here, more to the right. So accordingly, I'll move my correction also somewhere here. Let's listen. It sounds much better and the line looks pretty straight. Let's render and move to the second issue. This raises also note that highest, so I'll not go more than negative one semitone here. And I don't want to include this transition between the notes because it's fine. The band bends before the transition. This band is also shifted a little to the right from the center, which is why we're going to place this thought somewhere here. Let's compare. Excellent, It sounds right, and the harmonics are straight now. 48. Variable Time: As you already might guess from the name, the only difference between the variable pitch and the variable time is at the first one fixes the pitch issues and the second one fixes the timing issues. I used this module mostly for the dialogues because when it comes to music and my opinion, it's much easier to fix the timing and the DAW where we can use the bar scale I metronome. However, for this episode is example, we're trying to fix the timing and the acoustic guitars recording. Let's try to find the range of the wrong timing. It starts somewhere here and ends approximately here. The difference here with the previous module is that when the pitches bent, we can actually see it on the spectrogram. But when we have the timing issue, we can only guess. I'll start by adding 1 between this range right in the middle. And as I can here, there is not too big speeding, so I'll try to reduce the speed on, let's save 10%. So we'll have 90% here. Let's try. It's better. But I can still hear to an imbalance in this temple. Feels like the speed change to be more equal unlike the triangle we have. Now, let's try to flatten it a little by adding one more point here and spreading them to the beginning until the end. In this case, the speed will drop to 90% earlier and it'll be 90% throughout all the ranch. Great. This is much better. But I also want to try to lessen the transient sensitivity because this recording is already sharp enough. We get this sort of punch every time to Guitar Pick touches the string. Like her. 49. Measurements: And we're moving to the last section of the module pattern all which calls measurements. It contains only four tools. They are simple and some of them already covered earlier, so I decided to talk about them all together in one episode. Find similar meant to find similar pieces in the audio. We just select the piece and choose how similar the part we're looking for. For example, if we choose one, it will look for the piece similar 100% to our selection. And if 0.5, then 50%, Let's try. Here is a beautiful acquire. Let's say by any reason we need to find everything similar to the second part of this chord. I select it, maximize the value here, and click Find All. And it couldn't find anything. Although I'm sure that this quarter repeats at least twice through the audio. Why wasn't it found? It's because there are no copy paste parts here. And even though it was the same chord, it was sung twice and it's impossible to play or sing something 100% the same. What do I do now? I lowered the amount of similarity, and here they are. Now if I click somewhere to place the play head, the selections will be lost. So I click here, add markers, and now we have these regions. We click on them and they all sound the same. Perfect. Now I can process them altogether if I need. Just a reminder of what we talked about at the beginning of this course. We can enter the marker menu by pressing Option M or Alt M, or by clicking on this tool. Here we have access to all of our markers and regions. To place the single marker, we just press M and to mark the selection, we select the area and also press M. Here is the marker, and here's the region. We can do the whatever we'd like, including exporting and importing, which is very handy when sharing the project or transferring through different devices. And once we don't need these markers and regions, we can select all of them and then click Remove Selected. The spectrum tool. We can only find out any details about spectrogram, the entire audio or the selection only doesn't need much explanation. The left channel is white, the red is blue. And here we also have access to the extended representation settings. These are the same spectrogram settings that we already covered in the settings episode in the waveform stats module, we can find out the advanced waveforms specifications of our audio file. We cannot change anything here, but if we click on these play heads, we can jump to where it was detected. For example, I want to see the place with the lowest RMS level of the right channel. Here it is. Or the sample peak level of the left channel over here. 50. Practice: Welcome to the last episode of this course, which I decided to dedicate to practice. We're going to work on the recordings that include pretty much common issues. Of course, it doesn't include all the possible problems that are excess capable to solve, but I hope that managed to explain you the use of every module. I also tried to provide you with suitable audio examples for you to try to fix different problems with the use of different modules, just like I did throughout the course. The meaning of this episode is to face the real life example, the recording that has more than one type of of an issue and maybe take it a few tips about the casual workflow on isotopes are X. In this episode, we have a piece of the classical guitar masterclass. It contain is both dialogue and music, which we can tell by looking at the spectrogram. These are the dialogue parts. And this part has precise lines which are the harmonics of the musical tones. We can also tell that this recording is pretty noisy because we can see the orange gray fractions. The next very obvious thing here is this low-frequency hum, which is somewhere between a 100 hertz. This part is extremely loud. And if we slide to the very waveform view, we can see a lot of heavy clipping through the entire audio. This is the typical amateur recording that you'll face a lot if you do the noise cleaning for a living or maybe you're a videographer who doesn't have high-quality audio equipment yet. Let's try to listen to get more of an understanding of what's going inside this recording. Before we begin. We did first and then it's followed embody. So yes, most of the problems we heard now we could, and we can tell from the spectrogram in the waveform, which are static background noise. Dynamic noises such as random clicks, Russell, phone rings, and a few guitar squeaks, heavy clipping, low-frequency hum, unbalanced loudness. And in addition, I also heard that the right channel is a bit louder than the left one, especially from the dialogue parts. The panning should depend on what's going on the video because this audio is taken from a video recording. But in this case, I saw a video and I know that it's supposed to be in the center. After we have an idea of how problematic this audio is, it's time to ask yourself a good question. What's the result we want to and what we can achieve here? And how much time can we afford to spend on it? Let's imagine these two situations. In the first case, this is just a random low rate freelance job or the client needs the quickest turnaround possible for 30 minutes of his street blog recording. We'd rather quickly apply the essential automatic processing and send it to the client. Because if we are stuck in this recording for a few days, neither we, where the client is going to be happy. In the second case, it's a recording of a wedding speech and your best friend's wedding. There is no tight deadline and you'd probably want to do everything possible to make it sound as good as possible. You're going to go through every click, every smallest Russell and try different tools on it. We won't spend hours on this recording of this episode, but this audio file, just like any other, is available for you to download. So you can try achieving better results by applying your knowledge and spending more time on it. There are a few major problems with this recording. So why don't we ask for advice for the repair Assistant. I know IT advanced that these solutions won't be enough. But let's just see what's going on to suggest at detected a clipping and a noise. And here are the suggested options. The clip and spectral de-noise, D clip, voice denoise and game. The game here is to compensate the lowering after the voice denoise. And again D clip and spectral de-noise probably with the different settings. In general, it's correct. We have clipping, so it did suggest the clip and as well as the spectral de-noise for the background noise. We can render it right away or we can open it as a module chain. Here we can edit these modules in any way, change their settings, their positions, and so on. But this works great when you don't need to constantly change the settings of the modules. Besides, it's only if you need to process some selection with only one of these modules, you'd have to disable all the others, which is taking more time to compare it to using these modules separately. Since this is complex audio, we aren't going to use the module chain as well as the repair resistant. Let's start by making the audio more friendly tour years. Let's manually lower this part with applause and then use of the clip claim. Something like that. So it'll sound pretty equal compared to the main part. We can also create a smooth fade in with the use of the fate tool. We can also make the whole recording lower, but it won't reverse the clipping. The sign peaks will remain squared. We can hear these squares as the distortion. Let's use D clip to fix it. I'll use this suggest function for this one. And it shows us 0 decibel range because what we have here is an extreme clipping and there is no point to narrow the gates because even 0 decibel value will round the peaks. At least this is what our x things, but we'll still be listening to this audio little by little. So if we detect some additional clipping, we will come back to the clip and use more precise settings. Moving on and let's quickly get rid of this annoying coming with the D HM module. Learn, compare. Amazing. Let's render it. Let's now try to lower the static noise. I don't want to apply the same processing for the entire recording because some parts contain dialogues and some music which may require the use of different modules or settings. And I don't want to include the applause, so I choose the dialogue range between the applause and the guitar. I want to compare the spectral de-noise with the voice denoise here. Both are in the adaptive mode because I don't see a good noise range here to learn from. Besides the background noise here is not that stable for its quality removal with a single noise profile. And I'll do the two comparison options from the voice denoise. First using the surgical mode and second using the gentle mode. Played very well before we start. Not begin with. And then it's followed, embodied in silence. Playing very well. Before we start. Not begin. This is a really hard choice to make, but I think I prefer the spectral de-noise because it retains more energy in the voice in addition to these enchantment and synthesis functions. Although voice denoise provided us with a more soft voice and I would've probably chosen it if we had lesser noise. Well, let me also try dialogue Isolate in this piece. Made me fall. We started to play, begin. And then it's followed in bonding. In this phase of silence. I think this is a winner here. It preserves the originality of the voice the most. It sounds cleaner, but there are still some clicks, especially plenty of them during this pause. We can't just delete this part because it'll change the timing of this audio. Since this audio was taken from a video file, we cannot change the iteration in any way in order to not to ruin the synchronization, we can simply quiet it with the help of this clip gain week just smoothly lower this part somehow like this. So it doesn't become quiet all at once and accordingly doesn't have this weird feeling when you've veins. We could also use the game module for this task because although the clip game provides us with more flexibility, it takes us a little more time to apply it. Y, let's say we need to reduce the loudness at every pause between the phrases. If we use the clip game for it, it will require us to make at least three clicks to create this game triangle to set starting, ending. And the points, I'm not a big deal, right? Just a couple of seconds. But imagine how many times you need to do so throughout this 16 minute recording. If you remember at the very beginning of this course I mentioned the instant process function. It's a great alternative to manual gaining. We enable the instant process and choose here again, we opened the game module, set it to the way we like, let's say negative 15 decibels. Then we choose it. And here we go. We just select the part and it gets negative 15 decibels. Incidentally, this noise in this case is our ambience. And to make it even more static, we can process this pause with the D click on the heavy settings. I'm done. This is much better since we are already in D click. Let's also process the first and the second phases with it. Next we have a very noticeable high-frequency strike here. Let's try to reduce its energy with the EQ module. If we go lower, it'll make a voice of UDL. Besides, it'll also make this piece of the voice very different from the other parts of the phrase that we don't need. Let's also set a clip gain here to reduce a little from the word well, playing video. Here we can add a little since it sounds a bit quiet. But there's still a pretty obvious interruption in the word. Well, let's try to find its harmonics. I think this is a, let's now select all the harmonics. Use our spectral repair module to suppress them. Eubanks brokenness. We can still hear clipping in the following parts. Let me fall. We started to play. Before we go. Let's try to fight it individually. Before we always started to play the leaf. That's not the beginning. We did for sound. And BC. And then it's followed embodying in this phase. In this phase, in this phase of silence. In some parts of this distortion can not be reverted already. Though this part is much cleaner than the previous one. We can still hear this distracting HIS. Let's try to process it with the spectral de-noise. And then it's followed. Embodying. In this phase of silence. It's clean, but the voice is more distorted, val, it's not what we need. So let's try the Learn mode. In this case, the his aesthetic. So the learned function should work well. This piece is good to learn. And then it's followed. Embodying in this phase of silence. Great result. You've made me fall. We started to begin. Here's someone turning a page. I don't need that, so I simply lower the gain to the bottom, not at the beginning. But now the silence is to sudden. No problem because we have the ambiance match. Let's learn the single room tone, and then let's just render it here. Love the beginning. We did for sound almost perfect. It's just high frequencies here are still cutting it abruptly. But this isn't a tale of the voice, so let's simply cut it at the high end with the use of this low-pass filter. Let's just make it more radical. Let's say at four K. And let's render the ambiance match on this piece where we cut it. To begin. Begin. Great. We also do the same procedure here. And then it's followed embodying. And then this here is also some rustling that is happening. Let's try D Russell for a maximum reduction. Please see this better, but there are still some leftovers. So I think it's worth polishing it with the deconstruct and follow embodying in this phase. And here is the cell phone rings. It consists of many short blips, and it's pretty hard to select them one-by-one with the magic wand or even brush or lasso. I'll just use this time-frequency selection tool. Select this and this. These are the two most active harmonics here. And let me try to attenuate them with the spectral repair. In this phase. In this phase. Excellent. Likely the guitar part is pretty clean. There's nothing much what a hiss and the squeaks to get rid of. The HIS is the same as here in the open part. So we're going to learn here and simply render this whole part. Guitar playing is the main topic of this video. So we'd probably want to preserve as much of the original sound as possible. That's why we should be very careful with the threshold and the reduction levels to make sure we don't cache anything but noise. I'd preview first the output noise only. Great. I don't hear any actual sound of the guitar in the cancelled signal. So now we can confidently render him. Now let's work on these three major squeaks with the guitar denoise. We don't need to get rid of them completely, just making them a lot less noticeable unless sharp will go for the average settings here, because if we suppress a lot of squeaks, it would be Valerie audible, unpleasant volume depths. In case we needed to completely cut them, we'd better do this with the use of spectral repair, which is capable to replace removed information with the surrounding areas. But since we just need to lower the presence of the squeaks, it's easier to do it with the guitar denoise. Lastly, we can use the mixing tool to rebalance the pad. It's more to the right so we can decrease the presence of the right channel in the left output and the other way around to increase the presence of the left channel than the right output. Now it's time to use the leveler. Since we have both dialogues and music here, we can separate and process the recording with the two different optimization modes. This part will be processed in dialogue mode, and this part a music mode. Let's go for something like negative 20 decibels, which is including the dynamics of this audio will bring us closer to negative 14 ALK AFS integrated. We also need neither S reduction or breath control here, so we just turned them off and these options will remain default. Now we open the loudness control and we see that the integrated loudness doesn't reach negative 14 intergrated GFS yet, as well as the true peaks value is not what we need. This video will be posted on YouTube. So we set negative 14 elk AFS integrated and negative one decibel of a true p. We also less than the tolerance because if we have a big discrepancy in these values, we not be able to fit the YouTube requirements. So we set it to 0.5 integrated loudness unit, which should be fine. After we process the stats are not always updating automatically. We can just close it and check our new loudness specifications with the waveform stats tool. It's perfectly negative 14 GIFS now. But the true peak is still not even close to negative one decibel. But this is absolutely fine because as long as it's higher than negative one decibel were saved from possibly clipping. We would try to get it close to negative one if we worked on a musical track and we wanted to make it as loud as possible. But in this case, it works great for us. So this is pretty much it. As I said, working in R x is a lot of bulk combining the modules, experimenting and finding a balance between the time spent on the audio and the quality of the result.