 Hey, it's Anfa and you're watching AnfaVlog. In the previous episode, I've been talking about compression, the basics of compression, what is threshold, what is ratio, what is attack and release and what is knee and makeup gain. So today I'm going to do advanced compression. I'm going to talk about different techniques of using compression and different variations of compression. Follow me! So, I assume that you have watched the previous video and you know what the compressor is doing, basically. So today we're gonna make some advanced techniques. There are basically two kinds of categories for dividing compression, I would say. One is what you can do with the center compressor. So this is kind of different techniques of using compression and the other thing is different variations of the whole process. So you need a different processor, so you need a different plugin. I think that technically, well, you could have a plugin that has all the features that you would ever need for any variant of compression, but I don't know of a single one, but before, before I talk about all these different types of compression and compressors, I want to talk about one parameter that I didn't talk about in the previous video, which is kind of like, it's useful if you're doing advanced stuff and this video is for advanced stuff. So, I just have this my order session. I've got three helm instances, so we're gonna synthesize some sounds that we're gonna use for compression. I also have an audio track so I can record my voice and show you something else. I'm gonna just add a cough compressor. Come on, where are you? There you go. So, I haven't told you what detection is. So detection by default is set to RMS. You can set it to peak. Difference is, this is the function that the compressor uses to measure the input level, so the input loudness or volume. Peak level is extremely fast in response. It basically takes the highest value of any peak of the signal and takes that. RMS, however, is a more gentle, smoothed-out algorithm. It stands for root mean square and it's a way of calculating the effective power of an electric signal. And this, like, comes from the electric world where you just measure the effective power of an electric signal, just electric current going back and forth, alternating current, which is exactly what sound is in an analog circuit. In our speakers, in the microphones, the sound is just alternating current going back and forth. And here I have a program called JKmeter, which stands for jackmeter. And this program is currently wired, so my voice is running through it. So when I go, whoo, yeah! You can see there are two things happening in this meter. It measures RMS, which is effective power of the signal, root mean square, and peak level at the same time. The RMS level is denoted by the colorful bars lighting up, and you can see that it's reacting rather slowly. It's slowly rising up and then slowly falling down. You can also see that it is way lower from these white bars floating above it right there. And the white bars are the peak levels. However, the peak levels have some time that they stay in place after being incremented and then fall down. Because if this meter showed us the peak level without any decay time, it would be unreadable because it would change so fast. It would be going up and down, up and down. So we really care about the peak level in a time window. So the highest value of our signal in any given time window. And this is why it has some hold and decay. So when I just make a sound, push! You see it goes up or rather goes right. And then it stays there for a while and then it starts falling. This is decay. So it's going up. This is like measuring the top peak level. Then it's holding and then it's starting to decay. And that is made so it's easier to read the value because we don't really care about the intermediate peaks values. Because that doesn't tell us anything. We need the highest peak value in a time window, right? But the RMS constantly and it's falling constantly. And that's the nature of RMS. Of course it has some extra decay. So it's falling down a bit slower. It's falling down a bit slower than maybe it would for the sake of being more readable. But this is how RMS behaves. So when you use RMS measure measurements in your compressor, they are more smoothed out in time and basically slower than if you would use peak level metering. So there's a mathematical fall in formula. But what you need to know is that RMS basically smooths out the measurement. It has some time averaging, right? It's not instantaneous. So the basic rule of thumb is if you're using RMS and you feel your compressor isn't snapping enough, isn't reacting fast enough, you just switch this to peak and then you can catch the peaks much faster. However, that can lead to some maybe some distortion. So be careful to not have too short of an attack or release. Usually if you have too short release, it's a problem. All right, let's talk about different types of compression. You probably heard about limiting. And basically limiting is compression with a higher ratio. There is a definition that compression with a ratio higher than 10 to 1. So you see, we have this function, I'm going to turn down the knee so we can easily see the threshold which is here. If we make the ratio 10 to 1, something like that is considered limiting already. If you turn this 20 to 1, actually it goes to infinity after 20 to 1, but if you turn this to 20 to 1, this is considered brick wall limiting. The thing is you can actually use a compressor as a limiter if it has low enough attack and high enough ratio. However, limiters have some extra features. And this is where we branch out to like specialized types of compressor units as opposed to just using what you have, the basic compressor unit, in a different way. So I'm going to open up at a limiter called limiter. I'm going to make them stay on top so we can have both of these windows and we can compare them. So on the bottom, we have the compressor. On the top, we have the limiter. They have quite a different layout and you can see that the limiter doesn't have the graph and it doesn't have as much control as the compressor. However, it has some functions that the compressor doesn't have, which are very useful just for limiting. So we could use this compressor as a limiter. However, the limiter has few functions extra. One is look ahead. And look ahead is allowing the limiter to cut the peaks and not let them through. However, to do it more smoothly because the look ahead option is delaying the signal that goes through the compressor, so for the limiter. So if you delay it by 10 milliseconds, it's going to also have the attack 10 milliseconds. But it's going to start turning things down before the transient hits. So basically, like you have a transient here, right? And our regular compressor with attack at 10 milliseconds would start compressing after the transient and the transient will go through. With the limiter, when it has 10 milliseconds of look ahead, it would actually delay the whole thing by the 10 milliseconds start compressing here, right? And it would compress or rather it would turn the thing down. So the transient would be fully turned down. There would be no transient going through, but it would start compressing before the transient starts. The problem with this is that it delays the signal. So if you then want to mix back the limited signal with the non limited signal, you have phase shifting because the two signals are not aligned in time and the phasing issues occur. However, it's nice to make the limiting a little bit less noticeable when you can afford a bit of delay. And usually for mastering, this is not a big deal. This is also why in mastering, people are using lineier phase EQs because they also have to delay the whole signal to be able to line up the phases. So the total delay is the same for all frequencies, but well, they are mixed out. So you have delay. For mixing, this is a no go because if you have a delay and you mix this back to with something or for example, a different copy of the signal process differently, you're going to have phasing issues, which is going to destroy your mix. But in mastering, when you're just doing it on the stereo mix, you're not mixing it with anything else. You're applying 100% of the processing. It doesn't do anything. All right. So that's look ahead. It's delaying the signal and starting to compress before the transient hits because can delay it. So it's listening. Oh, click. Let's start the delay. Let's start a compression. It's lowering the gain. And then it's buffering the signal for whatever you tell it, like 10 milliseconds and then letting it out already limited. But it started fading it down before. Kind of weird. I'm not sure if I can really hear the results of look ahead. I didn't do any specific tests, like comparing the same settings of limiting on the same music, turning look ahead on and off or setting it to minimum and maximum comparing it to. I didn't do that. So it's not a crucial thing. The crucial thing is how much you limit to me. There is another great thing that the limiter has that the compressor doesn't, which is over sampling. Here it is. We can have two times three times and four times over sampling. What this does is help us prevent inter sample peaks go through the limiter. Because what are inter sample peaks where basically if you have two samples touching zero decibels to consecutive samples, like you have many samples in this digital signal, but two of them right to each other touch the zero decibels. The thing is the samples around that probably are below zero decibels. So the signal that will go through the thing is going to cross the zero decibel level. So actually we are clipping even though the samples don't exceed zero decibels. This is inter sample peaks with over sampling. The limiter internally is converting the sound to a higher sample rate. So if you have 48 kilohertz and you turn over sampling to four, it's going to actually make the signal 192 kilohertz. So it's going to insert additional samples at the function. And then it's going to see, oh, these are actually above zero decibels. We have to limit that. Without the over sampling, it would just see, okay, these samples didn't cross zero decibels. We have nothing to do there, right? If you're limiting at zero decibel threshold. So yeah, this is to catch the inter sample peaks. And from my experience with extremely loud signals, extremely like extremely high transients, you can still have the signal go above the limit. So for some applications, I actually use two limiters with over sampling and a bit of look ahead, one after the other. So when the first one didn't catch the peaks quite well, the other one will. And that's to ensure that we don't cause any clipping after the export. So that's over sampling. It's increasing the sound, the sampling ratio, the sampling rate within the compressor to make sure we really can accurately measure the loudness of this thing before we limit it. So we don't let the peaks through even if they are not detectable, if you just measure the peak level of the samples at the current sampling rate. The big difference also between the compressor and the limiter is that the limiter doesn't have the ratio control, because the ratio is just infinity. It's limiting as much as it cast to to make sure nothing crosses the limit. And the thing is with limiter, if you turn this thing down, it's not only lowering the threshold, it's actually not. It's it's arising the input gain. So the signal is going to still be limited at zero decibels, but you're going to add six decibels to it before the limiting happens. So this turning this down makes it louder. With the compressor, if you're turning down a threshold, you're making it quieter. And then you have to use the makeup gain to actually make it louder. There is no function graph in the limiter so it's not obvious what it's doing. But basically turning this knob down makes turning this knob down and this knob up at the same time, which is not obvious. And this is the compressor and the limiter operate in vastly different ways. Another difference between limiter and the compressor is that the compressor has many more features that the limiter doesn't. For example, you can change the attack time and release time. Well, you can change the release time, it's of course you can, but you can change the attack time other than affecting the look ahead. Because basically we don't want the limiter to let anything go through. If we want something to go through, we use a compressor because the compressor will allow us to let some stuff through by increasing the attack or making the ratio smaller. Limiter has to limit. It's not supposed to let anything through. Okay, so that's it. Okay, so now let's go back to the techniques. There is something called parallel compression, also called New York style compression. The thing is we are compressing the signal hard and then we're mixing it back with the uncompressed signal. So it's sad to have kind of best of both worlds that you have the punch from compression, but also you have the air and natural dynamics from the non-compressed signal. I often use that on a drum bus. So I mix all my drum instruments, kick, snare, hi-hat, crash everything into a single submix bus, and then I apply parallel compression on that. Sometimes I do this to individual drum sounds to fill them in, but mostly on the drum bus to make the drums feel bigger, feel heavier, feel more powerful. I know that it's rare for a compressor plugin to have a dry wet control like the cough compressor does, which is very handy because you can do parallel compression without using external buses. However, I know that many people are using plugins that don't have a dry wet control and they just create a separate bus called like parallel compression bus and they put a compressor there, make it crash it hard, make the output quieter so it doesn't overshadow the main sound, and then they just make aux sends. So they have their main kick sound and they send a copy of that to the parallel compression bus. That bus compresses the hell out of that but makes it quiet and then the both are mixed on the master bass, on the master bass, bass, boost, bass, weird words, Canadians rejoice. Another thing that we need a specialized variant of the compressor unit is sidechain compression. So I'm going to add a sidechain compressor and you can see compared to the regular compressor this one has more stuff. So we basically have some extra controls like the detection part is moved from here to here. We have our graph which is okay. I'm going to reset the settings on this one so we have the same thing so we can compare this easier. We have a sidechain thingy here which is we can use the internal signal path so when I go to pin out settings you can see that the sidechain compressor has four inputs. So the regular compressor has only two inputs because it's a stereo processor so we have left and right channel. The sidechain compressor has four inputs left right and then it has sidechain left and sidechain right which are called sidechain and sidechain two and because of great features of Ardor we can very easily plug something in. We can enable a sidechain input and then we have this thing that we can plug here and we can have any signal routed in here and the compressor can use a different unrelated signal to evaluate the level of the sound and compress things. This is famously used for example duck your pad synth when the kick hits or duck your bass when the kick hits. I'm going to do this today we'll synthesize some sounds and play around with compression using different techniques and see what can we do what sounds can we get and how fun it can be. However there's more to sidechain compressor because we don't have to use the external input and the cool thing is I'm going to disable this the cool thing is the sidechain compressor also has a filter module so right now we have the internals input internally split in two paths and the path one is unmodified it's only being compressed and the path two is going through the filter section so we can for example cut off the lows no bass and use that to evaluate our loudness to determine how much to compress stuff we can do interesting things because we can for example compress a kick drum ignoring the bass or ignoring the highs or compress a vocal ignoring the lows do weird stuff we have some configurations for two filters white band is no filtering the answer is a bell filter so this is the bell no so this is the bell filter yeah and we can like fertile filter one we can increase the gain and change the frequency so we have a peak we can make a fro or a peak we can move this around and we have a hypers filter here we can also increase the level of everything with it I have split which is just just the hypers filter the rumbler which is the opposite it has a low pass filter and a peak so these are different variations of filtering that you can apply to the sidechain signal without using it an external input we could do the same thing just setting up and another bus doing the filtering or equalizing and then rooting it through the sidechain input but a lot of things can be done right here so we don't have to so this is using the internal sidechaining and this is using the external sidechain so we have the second input and only after you so enable this thing you can actually hear the effects of rooting something external into these inputs otherwise this is going to be ignored right so you have to enable the sidechain input otherwise this is going to take effect and with this we are using these inputs right cool every special variant of compressor is the dsr which is basically a less feature for sidechain compressor which is using similar processing you can see we have a dsr configuration for the filtering and it's basically using that dsr is used to remove the sibilance from the voice signal or rather reduce the level sibilance are these sounds right they can be loud and annoying and i'm using dsr in my internal processing as i'm recording this video let me show you yeah this is my carla set up for the vocal processing and i'm using compressor then dsr and i'm using equalization and a gate all kinds of stuff because sibilance are a problem they are loud they are obnoxious and we don't need them we need to cut them down i hear so many professionally done mixes commercial release music that didn't use the dsr on the vocal and it just hurts because every time the vocalist goes ash your ears just explode once you hear this you cannot hear it but please remember to use a dsr on any vocal track and then decide how much you need to ds don't ever think that you can go without a dsr don't you dare however i sometimes use dsr for aggressive leads synthesizers or special effects synthesizers to cut down the highs when they are getting painful which is very useful also you can use them on snares to make interesting stuff with snares or hi-hats so i'm gonna insert a dsr cough dsr there we go this is a dsr it has a threshold like a compressor it has a ratio like a compressor makeup gain two has the detection and it has two modes wide band and split so the wide mode will process the whole signal the whole spectrum of the signal would just duck everything when the sibilance are loud the split signal is splitting the signal at this split frequency so it's basically a crossover it's then processing separately the low frequencies and the high frequencies and it will duck only the high frequencies when the sibilance occur and leave the low frequencies unaffected this might sound better in some circumstances but usually i just go with the wide band because it's changing the frequency balance of the signal and it usually doesn't sound natural when i use it so i i just go with the wide band here's a dsr demo i'm speaking saying stuff as you can hear my s's are exciting the dsr i'm gonna bypass it right now sick sack of kangaroo shoes now let's ds so we are currently in the wide band mode if i crash my voice with dsr ramping up the ratio up to 11 and going down of the threshold threshold it will cut down the whole thing so let's um not make it so brutal and when we just try to make the sr enhance our sound you can hear that it actually does if i disable it the sibilance are painful and when i enable it it's they are not as as big of a problem so we can oh what they do oh crap all right no not as much all right we can also demonstrate the split and wide sick sacks of sugar super soft super soft and splendid she there's also the laxity parameter and i think this is basically the attack and release control well if we make the laxity too small it might also distort the signal especially if we make the threshold low enough to chuck chum like bite the low frequency stuff it's gonna just act distortion especially if i make it wide band just you can use this you can you can abuse the dsr as another distortion unit which is always a good idea to have in your mind to distort things distortion makes things better so the laxity is basically the combined attack and release control and we can listen to what it does if i just make the ratio high the threshold low and then let's just change it i'm gonna do a little hi-hat can you hear the difference yeah it feels like it feels like the more laxity the slower that the sr reaction is so it lets a bit of the high frequency through and then it just ramps down the gain reduction it's not acting immediately and that would confirm why setting the laxity to minimum makes the sr distort distort the signal because the same thing happens with a compressor when we set the attack and release especially the release too short it's gonna just distort and yeah so we have a filter we can hype as the thing and do an extra peak usually about four five six kilohertz is where the sibilants are you can also listen to the sidechain signal with this one and then you will hear what the compressor underneath is using to determine the amount of gain reduction which is displayed here so yeah this is a very useful thing always put this on vocals always okay so i've put tons of different things this this is all the theory i'm gonna show you today and what about multiband compression huh well multiband compression is actually super simple once you understand compression and crossover filters oh well it's compression it's it's simple in theory it's not simple in use in the multiband compressor you basically have a crossover which is a bunch of filters that will cut your input signal into multiple signal paths based on the frequency then you have individual compressors for each band separately each band has its own threshold ratio makeup gain attack release knee and detection type each one also can be bypassed or even yeah funny so we have subband low band mid band and high band this particular multiband compression compressor calve multiband compressor is four band three band multiband compressors are more usual but four bands is whoa it's a lot i can do a lot of stuff i usually do use this to make the kick and the sub bass play better together inside a mix and i usually do that while mastering if i didn't do well enough job and mixing and i just don't want to go back to the mix but also it uh it adds specific coloration to the sound because it splits it into multiple bands and the filters have all phase shifting and such to a degree and then it's mixed back so it makes the sound feel a little bit different try it on yourself good job it's sober yeah you can also use this to have fun with automation because you could bait you know also use this tool to abuse your your very sound make it very very bad distort you don't know what to do distort the shit out of it let's do some practical things i'm gonna delete all these now i'm gonna synthesize some sounds let's make a kick here we have our kick awesome now let's play with the first with the compressor let's see what we can do without with so you can use a compressor to add transient to drums with using short attack and high ratio and then you just turn it up a lot you can also use parallel compression to make this less intrusive i think that sounds a bit better we have tons of gain reduction but this is because we're just letting through a tiny bit of the signal and then turning it down hallway but because we're turning it up 10 decibels almost 10 decibels afterwards we are really accentuating that click so that's a regular compressor okay now let's make a synth pad maybe so we're going to use the sidechain compressor to make the pad duck when the kick plays right all right that will suffice let's keep the sound simple so we can easily hear the effects of compression so i'm now going to add a sidechain compressor calve sidechain compressor and we're going to use the external input so first thing i'm going to go to pin out use sidechain input select my kick track and then root it in so now whatever the whenever the kick plays the signal the sound of the kick is going to come here now i need to enable the sidechain input and we already have sidechain compression this is sidechain compression basically let's add a bass i think my overall level might be a little bit high now let's see what we can do with a limiter so we can see when i turn down it's actually turning it up so this is what i taught you about so you're limiting but you also are making it louder i usually set up my track so the limiter hits it with the drums for mastering for putting out a mix i don't use any limiting i apply all the limiting in the mastering because i want to level balance the levels well and make use of my dynamic range the best for the whole release which is usually an album and i don't want to limit this earlier and then have to have unused headroom in my master or have to limit it again so this is it i'm going to just now record a vocal track and show you how the s-ing works it's shiny shiny shiny i'm gonna just loop this shiny shiny shiny so that's an extreme the s-ing you can listen to the sidechain that's hurt shiny so because it's compressing based on that filtered input it only reacts when the high frequencies are present shiny that's very useful for vocals we can also do the same thing with the sidechain compressor without using an additional input so i'm gonna use a dsr shiny mode shiny shiny shiny shiny shiny shiny shiny so yeah we can achieve roughly the same thing but with a sidechain compressor we usually have more control because we have separate attack release and all that stuff the dsr is kind of simplified down but it mostly does the job very well you don't usually need to reach out for a sidechain compressor to do the s-ing but you can all right that's it there's a lot of stuff in this video i hope you have wide in your wide in your perception of what compressors can do and what they are for there's a lot of stuff compression can achieve and it's very very ubiquitous in music production it's really important you can even use compressors to distort things but it's not completely frequency neutral because it doesn't react as much to high frequencies as low so that can be it can be interesting to spice up your your tones and distort them in a more weird way so that's all for today i hope you learned something if you have any questions about compression or ardor the open source digital audio workstation i'm using check out my other videos or ask the questions and comments i will punch you to videos if i have them answering your questions if not i'm gonna try to answer and if that is an idea for a separate video i'm gonna put this on my to-do list and then make a video later probably also big thanks to the patreon supporters there are two new patreon supporters so big thanks to artem ziransov ziranov and francesco ania ania also thanks to hu the case hu hu the case case i don't know thanks man because of your generous donations i can do this stuff and well life requires money to go on which is simple and i wish i can do these videos for the next five ten twenty years and teach you everything about music production and help the open source community grow because i see that there is not much like this and i think it's needed there is a community but we need to see what the tools we have can do see who we are so i guess more interviews with open source audio developers and musicians will come to life as the first interview i did with mark mac curry the developer of ziransov effects and the creator of zinfusion which is awesome this video is a bit longer than i thought it's gonna be but well this is what you got i am working on a dedicated course for ardor because i see there is a need for a solid guide to teach basics and the fundamentals and then also not the fundamentals but do it in the organized way because people are coming and say well how can i use ardor well i have videos where i use ardor like this one but just diving in is is hard because this program is is big and it's complicated so i'm going to be working on videos to explain that to you in a separate series because i want this to also be you know just focused only on ardor no synthesizers just clear it so so it's just ardor so anyway if you like what i do and you think this is important and you want to help me out with this you can support me on patreon for as little as one dollar a month also you can support me by buying my newest album and bandcamp which is available for free dollars or if you become a patron you get a code to get this album for free as a bonus i will upload this session there's not super much fancy stuff but you can play around with it i guess and that's it well i'm set talking to for way too long right now okay see you in the next video bye