WINDOWS VERSION is now uploaded. PLEASE NOTE: if it wasn't clear from the video, my plugin does not magically undo ALL kinds of distortion. It is only a demonstration plug to show how, when you saturate something, you are preserving the details of the music despite the distortion and peak reduction, whereas when you clip, you simply delete your audio, loosing all information over the threshold. I think my plugin demonstrates this well but don't get carried away and think you can reverse all kinds of distortion with this, you can't. Feel free to play about with it but the reverser side really is a kind of trick for demo purposes only and I'm unsure you will get much genuine value from it. That said, the tanh saturation on the left of the plugin is very pleasant and useable, and if you can oversample it (like you can do in reaper for example) it will sound even sweeter..... ALL nonlinear processing is subject to some degree of aliasing and should be oversampled in the best case... 4x is fine but more than 8x is totally unnecessary (note to self: update the plugin with a second order antiderivative so it sounds great without OS....).
I think the whole point of clipping is controlling the crest factor. It's not just loudness, but hitting the right compressors/limiters with a more uniform signal, so it's not like choking on a couple of transients and completely missing others.
@@APMastering it's a youtube tutorial problem or information availability problem, not quite a clipping problem itself, plus it also depends on a stylistic choice and genre itself. I love me some soft clipped or tanh kicks and snares.
man I think I've been living under a rock.... people actually CLIP their masters????? I've been working for years in audio engineering and it sounds crazy to me people do this on purpose lol, as if squishing everything with a Waves L2 wasn't bad enough already. Maybe in the Rock and Metal circles I usually work on it isn't part of the culture to clip things and its more of an electronic music thing, or maybe I just got lucky, still great video man! Blew my mind seeing things being de-saturated!
I think this is the sort of thing that Airwindows console plugins are doing, doing some sort of distortion on each console channel and undoing it as one on the master.
You should have a chat/collab with panorama mixing & mastering on the topic of clipping. You lads both very knowledgable on the topic and have opposite takes - would make for good content/discussion. Moreover, you both have a catalogue of work behind you so there is established experience in the industry behind the opinion.
@@APMastering ive seen his stuff and i think you really do have a lot to talk about. to be honest, i used to have a very similar approach to his when it comes to getting rid of random peaks - i just clipped them. after watching your stuff which opened my eyes i got back to the traditional levelling approach.
@@SAFAMASTR the difference was barely audible. no matter how you do it, getting rid of random peaks makes the whole thing sound more... precise. that being said it's much more important when you're actually using a limiter in the end, something i almost never do. (or at least not in the way most engineers do it)
@@LeChapeauMusic i get this. Especially if you're mixing with high unrestricted headroom. But getting sound even close to the recommended streaming loudness of -13/-14 lufs usually requires some limiting. And the last thing you want to do is to go below, since they might boost and limit it for you, badly... And that's where clippers can be extremely helpful and actually get rid of noticeable artefacts (done usually by the limiter), not introduce them.
In the kick example - 100 samples is only 2 ms - it will be audible when you play a 100 sample/2 ms white noise sample in silence, but the same 2ms in a mix is not really audible at all - even less so when you include the fact that the initial transient of a kick is usually high frequency heavy, and is on beat with other instruments which helps mask the white noise. I use clippers on drums and limiters on tonal instruments, and the tradeoff is worth it - the final limiter barely works more than 2-3 dbs before I get to stupid loud (-6 lufs or so) The sad fact is clients always A/B their mixes and masters against other stuff they hear on Spotify. If its not similarly loud to other releases, then you get the blame. While I agree that there are some terrible examples of shoddy production, this should not reflect on the potential of the tools. Looking at badly deformed nail on the wall and blaming the hammer isnt the direction we should be taking.
Like the content and insights. A key difference with tanh vs clipping is that with clipping you can affect a transient with no change to samples below the threshold. But with tanh you must accept the gradual onset that will modify samples below the threshold. Both result in a distorted signal, and which one is more noticeable is program and threshold dependent. Assuming undo-ability is not desired then the choice comes down to what your sound goals are.
i agree but the nonlinearity of tanh doesn't really bother me until the signal is real hot. pretty sure i could run a string concerto through tanh and nobody would notice given a bit of headroom
There's a deliciously unsavory irony to be found in listening to a comparison of clipped vs non-clipped signals through the RUclips smoosh-filter. Honestly, I could hear no difference in that snare example. BUT I 100% agree with your conclusions.
@@APMastering Listen at 2x? I tried it once and it made my soul hurt. No, I listen at 1x always. BUT I am in my seventh decade now, so if the juice is flowing above 10kHz it flows without me.
It's not just about loudness, it's about not loosing punch while making it loud. And the signal is going to be relatively louder than other songs and will be percieved as better. It will even sound louder when turned down by a streaming platform. Even phase rotated, it will still be louder. There is nothing a listener hates more than having to readjust the volume for every song. Increasing the volume of a quiet song in order to match loudness of the soft-clipped song is going to cause amplifier clipping. And distortion can be undone no more than any process can be undone with an undo button - if you have the original information or the information about the process applied to it. You can also undo clipping if you lower the volume so it's not clipping anymore.
"And the signal is going to be relatively louder than other songs and will be percieved as better. " No it won't be. If it is loud but sucks, it will be deemed a bad mix and the listener will turn it down. Average listeners might not be able to talk anything technical but they know a bad mix when they hear it. "It will even sound louder when turned down by a streaming platform." This is completely false. "There is nothing a listener hates more than having to readjust the volume for every song. " Do you have evidence? "Increasing the volume of a quiet song in order to match loudnes of the soft-clipped song is going to cause amplifier clipping." This is wrong. Speakers and amps can typically deliver double the peak power as sustained power. A clipped song decreases the crest factor, meaning the song has less peaks and much hotter sustained power with less loss through kinetic energy which puts more strain on the amp and can overheat the voice coils. "You can also undo clipping if you lower the volume so it's not clipping anymore." No you can't if it is committed. The point in this video is that saturation maintains detail AFTER BOUNCING. Whereas clipping destroys it. Hope you can learn something from my response.
@@APMastering You have an idealistic view of the world. No, average listener can't tell a fart from bass guitar, let alone recognize a good mix. Loud songs do not suck by default, this a totaly misplaced argument. Louder is percieved as better even by many mix engineers, you bett an average listener is going to percieve it as better sounding. "This is completely false" - Try streaming normalized songs, there will be difference in percieved loudness. "Do you have evidence?" - Funny how I have to prove things in comments, yet you don't prove your claims in your video. I and most people I know, hate readjusting volume, and if anecdote is not enough for you, it costs energy and it's a distraction to have to go to your amp, or look for your phone or the remote, when you just want to enjoy music, even worse, if you're busy doing something else while listening like most people do. In practice, speakers are typically not going to be fed unaltered squarewave unless the clipping is occuring internally. Your arguments are really rationalisations of your biases. A commited song can be replaced with the unclipped version of the song. Your point is no different, you need to store information either about the original signal, or about the process that's been applied to it. I hope you apply logics and realize that you aren't teaching, you're just expressing your point of view.
@@duncan.o-vic I would say, that a average listener can say, if a song is mixed well or not. That for sure. Loud songs, which are pushed need to be very well EQ´d. There are a lot of things which go into having a great sounding, loud mix. Starting with the low end. But one point I want to say as well. I sure as hell hate, when I need to fix my volume while listening to a playlist lol But I also do listen without loudness normalization, which one CAN definitely turn on. This will help to avoid that, but you also then not hear the mix, as it was intended.
I knew. :) Distortion is, of course, a "linear" process (minus the sample/bit resolution reduction). I used to do this years ago with an ancient distortion plugin where I could make a custom curve. I could 'magically' clean up old recordings to a certain extent, except clipped audio. I thought most sound engineers knew this, though!
not sure what you mean with linear but undoing distortion is pretty esoteric afaik, cool that you were doing that ages ago, i never was so successful at it in restoration work
@@APMastering What I mean is, it can be undone. It doesn't introduce time-based distortion like compressors. i.e. You cannot expand a compressed signal back to the exact original, but you can undo waveform distortion if you know the curve used to distort the waveform (without clipping, of course). EDIT: I know "linear" probably isn't the correct term to use when it comes to waveform distortion. Can't think of the correct term to use.
This is an excellent approach to showcase the reverse function to a broad audience. As I can see, it is not applicable for 32-bit (float) signals, as we lose precision with each iteration, but the signal samples at 16-bit/24-bit depth are considered integers, and everything should work just fine. Anyway, I still suspect there can be corner cases. I like your channel!
This is the kind of stuff that every real Engineer should know. Thanks for all you have done for us and you help us out tremendously. Great videos explaining everything.
@@APMasteringis there any reliable way to learn this information from a track? I have been told in the last Waves makes plugins that can undo brickwall limiting and compression and restore dynamic range to modern releases
So, a certain TYPE of actual distortion (in this case algo of hyperbolic tangent, which compresses intensely the original values of the sound into a hypo-tan set of numbers). This can be undone by the inverse function. Makes sense. Clip is dangerous because the original harmonic content has been wiped out and no longer available. It may make sense to saturate gently, but having clipping usually is going to be quite regrettable, especially if it is printed permanently into original multi-track sound. I'd agree that "saturation" is not really clipping per se. It has similarities but it's not really the same.
Hence why music production is not just objective it is also subjective hence why the tools are used and abused for creative uses not just for the technical aspects. Best of both worlds.
@@APMasteringbut digital information loss can be a desirable too. See bitcrushing, samplerate reduction, deliberate aliasing, low bandwidth codec plugins for things people use
This is kind of operating under the presumption that clipping is the only process contributing to loudness for some people. Typically clipping with K-clip is just the final process after saturation, compression etc to control those 'every now and then' unwanted transient peaks to prevent higher peaks in up stream bussing at least for me. K-Clip might not even meet the strict definition of clipping you've laid out, so saturation is really what's putting the work in loudness wise for me. I never want the clipping to be audible or adding any kind of unintended colouration, I just want mixes with the kind of marriage of punch and density expected in a given genre.
I agree that mixing loud is MUCH more than clipping. In fact you can mix loud completely without clipping. I think kclip is incorrectly named. It's a multi band distortion plugin.
I think you may be triggering youtubers who rely on the esoteric way to describe plugins. I appreciate it though. the producer crowd sometimes feels like they use vague language on purpose to make themselves sound more skilled. Whereas the music performance youtubers seem to be the opposite. Play this scale in this way with ghost notes to get a Swing saxophone effect. Versus "use my effects chain to make dope beats". It's infuriating, honestly.
maybe but i think there's this other dude with 30 extremely long videos on clipping that built a following and people think they need to use 100 clippers in their mixes
@saxmanash Selling sample packs is something I can still understand, selling effect rack presets - sure, why not. But nowadays these people are also selling midi chord packs, and apparently getting sales. This is something I will never understand, with a bit of reading into music theory you can construct any chord you want, why are people buying digital note values… Are these the same people that buy NFTs? 😛
almost nothing. i genuinely can't think of a single valid use case in music apart from maybe synthesising a square wave from a sine wave but ironically it's more difficult to create a sine source than a square source anyway
i'm definitely also more a fan of softclipping than hardclipping but i wouldn't go as far as to say that hardclipping is never useful and just makes things worse. the harder you clip something the more direct and close it sounds. you also have an easier time controlling the peaks and get stronger harmonics if you clip harder. but there are a lot of sounds that really don't like being clipped too hard, like vocals. clipping is also useful before compression to make the compressor react more evenly to a signal, and it's useful after compression to get an even sounding distortion, but that is true for both soft- and hardclipping
I think we have slightly different approaches... maybe im more old school but I never use clipping. I feel like there are many other processes which I am happier with than clipping that achieve similar processes, or I find the processes unnecessary.
You videos offer very high value. I have always been dubious of clipping. And came to that conclusion after rigging up a couple of anti-parallel germanium diodes in the signal path of my mastering chain many years ago. Indeed the only audio that should be subjected to such shenanigans is the signal from an electric guitar...! OK...perhaps some synth stuff as well... ;) Just not mix busses or masters!
Unpopular opinion: If it sounds good it sounds good. If those bits sounded bad then delete them. Genre dependet i think resloution is overrated. I dont care how the effects are done, as long as it serves the music. Sometimes there is a use case for hard clipping before a process like compression regardless if there is a phase rotation after. Maybe soft clipping is the move, maybe not. Ears are the final judge.
There was a "clip restoration" function in the old Adobe Audition - I had some success with it on badly recorded / distorted audio. Didn't suit everything, but when it worked, it worked very well, and saved somebody's session.
Hey nice video mate! I don't see however why you treat information loss as a bad thing. While mathematically you're right, saturation does reduce the volume of the information, and nobody does atanh functions on their music. Clipping just lowers it to zero. There isn't actually that much of a difference, is there? About the PA systems, I guess that means you can't play square waves either? I don't find "damages the PA system" a convincing argument unless you're mixing specifically for speaker longevity. It's a side effect. And the argument about clipping mixes applies to sat as well. This reads to me as a bit dismissive of clipping in general? I think if it sounds good it is good, and while I normally think saturation does sound better, there's a place for hard clipping. Specifically, one advantage it has is that the parts of the audio that aren't clipped aren't also altered in any way - they're just turned up. This is very useful if you just want to turn down a peak.
People use clipping and distortion for different reasons, distortion is changing the sonic character, clipping cuts off the inaudible peaks, so they won’t trigger your bus compressor. These are not equal steps in the prodction process, both have their valid place in the workflow. And why anyone would want to reverse the distortion made by a specific plugin? In case you’ve lost the project file? Otherwise you just go back and export it with less distortion…
@@APMastering okay, and the other half? That saturation is an artistic choice, while clipping is a utilitary thing, that helps your bus compressor’s work? I think you’re forcing this idea to fit in your concept about “clippers are a scam”, but it doesn’t hold up. Cutting off inaudible and unimportant transients on a drum hit is a real thing
@@SenorTropiCatbruh if you do this with "free" plugins then a 200 dollar clipper is a scam I don't care what creative purpose you give something That's just because you are tryna cope with spending 200 bucks on a piece of software that is not worth that amount of money Regardless, stop spending money on plugin they don't make you better
@@SenorTropiCat I think clipping can be creative too, but I'm into harsh noise and other extreme genres as much as I'm into more conventional styles. I feel like there's a dangerous dogma around the idea of resolution and quality that -- although it might be more based in data -- is ***similar*** to how audiophiles think of audio. If it works for the genre it works. If hard clipping a snare sounds better than soft clipping a snare I'll happily delete those bits and never look back. One of the most useful things I learned in my undergrad studies was that there's no such thing as a bad sound, just a sound that's wrong for the context. I feel like those nuances get lost in these discussions.
@@Notinserviceij who said i’ve spent 200 dollars on a clipper? 😹 I’m using the clipper in the TDR Labs limiter, a plugin that I would use otherwise too… Even without the clipper part.
Hi 👋 😂 I was “that guy” in the last video. You’re not necessarily wrong here (nice to hear someone talk about the reverse saturation option. I discovered this when playing with waveshapers thought it would be a cool video idea. I like to put processing In-between for weird sound design experiments… anyway) there’s a couple of things I disagree with again. Hard clipping is actually cleaner than Saturation. Why? Because your signal is spending less time in a “non-linear environment” Intermodulation can be reduced using hard clipping. At least it sounds cleaner on complex waveforms like a mix. 100 samples out of 44100 per second is still very little. The question is do you hear it? If the answer is yes, you’re doing it wrong. I do often use “softclipping” as it is “softer/less instant” but at the end of the day you could argue this is also reversible. But it really isn’t because of the signal is loud enough we’re limited by our bit depth to realistically restore that almost.almost.almost clipped information. and realistically the only difference between hard clipping and saturation is how steep is the non-linearity. (Assuming symmetrical +&-) Also aren’t all speakers high passed and low passed anyway? Meaning any “prolonged energy” would be offset by the phase shift in the speaker itself? This destroying speakers argument is more to do with sheer volume than the shape of the waveform. …. God this is a deep topic. Appreciate seeing a different point of view but I still think clipping and “soft clipping” are very useful tools. Also no shade here. I appreciate the video. Nice to nerd out with fellow nerds.
@@dodgingrain3695 yes that’s the essence of what I’m saying about having less perceivable intermodulation with a clipper compared to a saturator. Meaning it can unintuitively sound cleaner. Clippers get a bad rep but they’re actually the most transparent peak control tool out there. When used properly.
The whole cant undo information loss thing makes me want you to base part 3 around AI tools meant to restore information. sony DSEE HX, Izotope Spectral Recovery, voice fixer, heck even the old low bitrate restoration trick of cooying the track shifting it up an octave and EQing everything below the cutoff on the real thing away Izotope declipper i have had make some good results Audacity not so much but maybe its for more specific issues
although I'm a mastering engineer I dont tend to do much actual restoration work, so this isnt really my strongest area. However, although I have used izotope rx in the past to sort of improve some really terrible stuff, I don't think any of that stuff is a substitute to getting it right in the recording, as none of it really sounds that good. I would be interested to interview someone who does actually restore historic recordings etc but this is not my thing at all.
i'm not wrong about all pass on radio, look it up. i'm also not wrong about speakers, if i were all of those reliable sources would be wrong too obviously
So in terms of damaging speaker cones and what not, essentially the idea is that hard-clipping introduces DC offset to the signal (which basically means that the signal has more positive or negative values overall). But doesn’t heavy saturation also introduce some amount of DC offset? Especially when applied to non-symmetrical signals? So, I don’t think the kind of distortion really matters in terms of DC offset. If the PA system is handling audio properly, and the audio signal is being High Passed to cut out the DC, either in the actual recording, board, or in the drivers, then the DC offset thing is kind of a non-existent issue. Audio signals aren’t even truely “flattened” when being clipped anyway, because of the nyquist limit. There’s still some actual wiggle in the signal (recall: true peak) So correct me if I’m wrong, because I’m mostly going off of intuition, but I believe the issue only really arises when the DRIVERS themselves are being clipped. THAT’s where the issue comes into play. And that’s not an issue of the audio itself, but rather an issue of the input volume simply being too loud going into the speaker. But if the drivers aren’t being clipped, an audio signal with a clipper on it isn’t truely “clipping” bc of the Nyquist limit (hard LP filter). And any dangerous amount of DC offset would be mitigated with a HP filter.
sorry for short reply but has nothing to do with dc offset in the normal sense, unless by dc offer you mean the clipper top and then that's not really the correct term. it's overheating because of power without movement
@@APMastering no worries. My point is that “clipping” on a sample level is not actually power without movement, because of true peak: the samples don’t reflect the actual physical “movement” of the speaker cones, but are something of a “guide” for the signal.
This might be missing thing from digital audio i've been looking for you got here. I've had a problem with getting drum compression sound the way i want. I always seem to not be able to get that smoother fat tone and end up with eather too much spiky transient or pumping. Drums sounds i like are from the analog days and tape compression is just like that, it rounds the sound. I think the flat waveform tops are what makes also human ears tired when listening some loudnesswar stuff with headphones you feel this constant push against your ears but if you listen to something older that has tape compression and tupe saturation in it, it can be loud but will still sound smooth and more pleasing to the ears. If your plugin is a way to do a digital version of tapecompression like it seems to be it's really needed. If you have it also really clean without aliasing or similar wierd harmonics (i wrote this in the half point of the video)
does that mean if other plugins had this opposite algo, they would also be able to undo ? any strange cool sounds emerge by putting other effx between the sat & the reverse ?
More click bait, just sad, and sad that people believe this guy. Again, I don't think you understand the use case for a clipper. Obviously a clipper tends to make a drum hit feel like it has less punch but it does make it louder because you're TURNING UP THE GAIN. You could maybe argue for a more feeble sound but it won't be quieter. Nobody clips a signal and then turns it down either, ROFL. Also referencing Ethan isn't helping your case. With speakers I think your also confusing a square wave with clipping, they are not the same. Square waves are used all the time in music especially as a bass sound and it doesn't inherently destroy tweeters. Speaker manufacturers are not worried about squared off waves, they are worried about excessive high voltage. Case in point at 11:35 "its sustained power", not squared off waves. The "unknown reference point" is NOT the users volume knob, its the previous song. Distortion, regardless of the type and regardless of the chain in lufs is perceived as louder even if it isn't, clipping is just another distortion. Seems like your best argument is that someone that doesn't know how to mix well won't be able to get a mix as loud as someone that does know how to mix..... which has nothing to do with clippers. So much other stuff is wrong in this video, there just isn't time to comment on it all, but hey one last one, the notes from the book at 18:05 are essentially telling you to use hard clipping and avoid any kind of transfer function that isn't hard clipping. Also that book is taking the standpoint of clipping only from a utilitarian view and not as a potential creative tool.
Ive been waiting for this to become a thing, since the latest wave of truly brain-erasing AI assisted noise reduction and de-reverb plugins started surfacing. Let's say,you've got a spanish guitar thats badly clipping. If If a human can visualize what it sounds like without the distortion,it should be possible to use generative AI to pretty faithfully reproduce a clip without the distortion.Amazing time in DSP processing for audio.
yes! but the way I see it, part of the greatness of great recordings was the time, place, coincidental stuff that all came together. AI will make everything great and then relatively speaking nothing will be great any more.
@@APMastering Dont disagree, but that is neither here nor there, not unlike how an AI generated vocal stem will never be 100% exactly like the source. Damn close,yes, perfect, no. The new paradigm, that if a human mind can visualize what an undistorted verson of problem audio would sound like, than a neural network can, too, and so, this should be possible. (popcorn!) haha
@@APMastering If you would have asked me 5 years ago if this would be possible one day soon,Id have to say no way. Imagine that just a few years ago it really looked like we were at the limit of what could be done and how of how an audio signal could be manipulated digitally or otherwise. Im looking forward to what this means for restoring old damaged recordings, forensic audio, moreso than music,per se. In the future, i believe music producers will no longer play musical instruments, nor have or have to have any technical skills moreso than any other person. if you want to make a song, just think it. Like most aspects of ai, this both fasinates, and saddens/terrifies me. Someone wake me up when ai does something tho help, not ruin society at large. :) Still though, may you live in interesting times!
sorry, but after watching this video all I think is... my dude got gaslit into believing that we clip entire master in electronic music or hip-hop or smth with +15db clipper or smth... or we clip the everliving crap out of drums, information loss, blah blah... great stuff. again, skill issue from rock musicians, mixing and mastering engineers trying to chase the high from high lufs. dudes just got sent 20 years ago back into loudness wars insecurities. so sorry to people that have to go through this.
Just a question The saturation in your plugin is a achieved by a linear process. analog saturators are based upon nonlinear processes that cannot be described by the means of transfer functions. In that case, it is not possible to reverse it. Am I right?
A linear process is something which can be described as output = SomeFunction(input), as in this case. A nonlinear one can't... However, I don't know how nonlinear are the equations of saturation processes inside real hardware. In my opinion this video should be called "plugin saturators can be reversed". Good video btw
@@veinsbandofficial6221 again i dont agree with what you say linear means, it's a well defined term. But sure, hardware likely has more complex distortion characteristics than digital tanh
@@danyavilaoficial “clipping is bad” is the bullshit. It’s an outdated narrow minded dogmatic philosophy. Any tool can be used for harm and good and the audio engineering community should embrace clipping as the useful tool it is imo.
Hard clipping can potentially create a DC offset because the straight line of a square wave is DC. A DC offset can potentially mess up your amp and speaker biasing, causing damage, heat, or extra distortion. However, this is only theoretical, as all of them have so-called coupling capacitors that block DC. So, I don't think that square waves will damage your speakers on their own, it's common practice to use square waves to calibrate amps, speakers, tape recorders, and so on. The dangerous clipping they are referring to occurs when you feed your signal to the speaker or amp so hot that it exceeds the power rating of the system, overly loud source material only adds to that.
this is wrong. Although the top of a square wave is static, it's not electronically considered DC since it is not 0Hz... let's say the square wave is 1kHz. this is higher than 0hz therefore it's not DC. being reductionist about just the top portion is not how people normally discuss this
@@APMasteringTo achieve the flat line of a square wave, there must be a DC component present. Even if the wave is 1 kHz or higher, it still contains that component. You can test this yourself: generate a 1 kHz square wave and apply a 1 Hz high-pass filter on top. Then, observe the oscilloscope to see how the wave changes. The flat line will drop even if the filter never affected the fundamental frequency, because it removed the DC component.
@@eliashowe7115This is not right. Clipping can essentially induce DC, but not because of the flat line. It's because the energy imbalance in the top/bottom swing of the waveform becomes exhaggerated, depending on the material. You can usually see some DC on an analyzer after clipping.
Great video, it's refreshing and rare to see people exploring how things actually work and providing a spring board for constructive discussion! Does the same information loss apply to hard knee limiting where it seems to flatten a peak instantaneously before having a gradual release? Is this the same as a hard cliper on the onset while the gradual release is mitigate some distortion for cutting into the waveform or is it somehow compressing the dynamic range instanteously in a different way? Does that mean a soft knee limiter is functioning the same as a tanh function on its onset and by this measure would you only ever personally use a soft knee limiter?
@@APMastering Keep us updated on your findings, it would be silly to be avoiding information loss with 1 process only to be unknowingly doing it with another. I'd love to see a video on the process of building these plugins, including the research/circuit analysis, coding and UI/UX if you find the time.
Can you do a video on newfangled saturate. Its doing something different and I'd love to know the downsides vecause it sound freakin awsome and loud but it kills the bass a bit. Its a clipper so the name is kind of misleading
Tanh is too soft for mastering use unless you're happy to distort low level samples, which generally isn't desirable across a whole mix. The term "soft-clipping" is generally used for saturation curves that are much harder. Closer to full on clipping, while retaining a soft knee. For example: x/(1+x^8)^(1/8)
sure my free plug wasn't aimed at mastering engineers. As a mastering engineer I don't typically use this kind of processing anyway. Having a steeper function has its uses maybe as a part of a limiter design but I don't see how it is so useful in mixing personally.... I guess I'm just not cool enough to want to nuke the dynamics of my music unlike many of the people who commented on my last video. Anyway I tried out your math.... channelData[channel][sample] = channelData[channel][sample] / pow(1 + pow(channelData[channel][sample], 8), 1.0 / 8); imgur.com/a/1r0XHZs
@@APMastering K. I just want people to know they don't have to jump all the way to tanh if they want to avoid hard-clipping. For me hard-clipping is full avoid, but a slightly soft knee does have its uses, especially on drums. Once you get to tanh I would call that saturation, not soft-clipping, but all are part of the same spectrum, with only hard-clipping being strictly defined.
ok so your point is that soft clipping and saturation exist as separate terms, based on knee, and although there's overlap you are willing to say that soft clipping exists in between clipping, and that saturation definitionally doesnt get close enough, therefore the term soft clipping is required. What I'm saying is, saturation is everything of this sort which is not clipping, and there is no room for the third term, soft clipping. We understand each other. But we disagree. I will stick to my guns as I'm not convinced that saturation is inconsistent with what you call "soft clipping". I'm open to having my mind changed but I see no reason right now to change it based on mere popularity of the term. I'm sure I could demonstrate electronically a type of saturation which comes close to clipping, like for example saturation of an op amp based amplifier which reaches the power rails.
Question: I noticed that if I run a mix through a dithering plugin like the stock dither that comes with Cubase, using 16bit dither, I can drive the signal, it appears to "clip" (reaching 0db) but with no significant artifacts in the signal being introduced. What could be happening here? Is a dithering plugin acting as a "clipper" somehow? I dont use this of course - I was just curious after messing about with various dithering plugins.
it might have a built in clipping function. im not familiar with this specific plugin since I dont use cubase (well I haven't done for around 20 years).
@@APMastering If I understand them correctly, Airwindows' Consoles work in this way. Saturating at the channel stage and de-saturating at the bus stage. The de-saturation isn't a perfect null because of how the tracks sum and play together, giving us this sense of "glue." Feels like things are saturating and working together instead of individually saturating and still playing back separately. Chris would obviously be able to give you a better explanation though!
@@APMastering and now I think about it, you and Chris should consider collaborating. You two have similar messages in your videos and products. You giving his plugins some much needed love in the GUI side of things (and possibly AAX support) and combining your processing philosophies would be killer. APWindows could change the game
would you debate your viewers on this topic? I feel like, again, this is quite an interesting topic because of a lot of information missing from the perspective of both hip-hop and electronic music. for example, there is only a handful of people that clip individual elements like that. that one baphometrix video doesn't actually show how it actually works.
Are you planning to expand this plugin to have more trransfer and... untransfer (ig just mathematically inverse)... curves? Seems really cool. I am curious what your take on De-Clippers is, like Izotope or Adobe or FL Studio. Obviously the original it can't be recovered, however the ability to "redraw" the wave definitely seems even more interesting after seeing how your plugin works in this vid
de clipping is impossible. you can only interpolate. if you want more drive for the saturation, just boost the level up before the plugin. you can boost 1000db and the output will still not exceed 0db 😎
@@APMastering I meant outside of the saturation context - more like for dialogue and stuff. I guess the "interpolate" you mention is the "redrawing the wave" with tools like the RX, which is what I was curious about your take on. I understand de-clipping in the literal sense is impossible, but what I'm more so trying to ask is about your opinions in regard to tools like the RX De-clip, Adobe's De-clip, the FL Studio De-clip is - outside the context of mixing, saturation, and distortion - for things like when a voice actor might get too loud on a certain word or line and clip the analog input into a recording, etc. Trust me, I know there's limits to tools like that, but I'm interested on your feelings about tools like that and your insight into how they might work due to your level of in depth explanations.
interesting concept and plugin. my question is though, does the plugin only reverse the type of saturation thats also produced by this same plugin or will it effectively eliminate other sat types/shapes?
Another very simple and useful tool and a super educational video, fantastic work! Just got the windows download. I appreciate all your free creations and while I'm not supporting you financially, I'm happy to support this content by word of mouth. How are you doing mentally? I'm imagining there's lot of brand new stress and anxiety that you're exposed to lately, with every vid being a controversy of some sort lol. Take some time off from responding to hate comments once in a while, it'll screw you up eventually and you'll end up responding to everyone with same energy. No good! So stay humble and healthy! :D
I will try your plugin on my song where i don't like the spikes of the drums attack that sounded bad with limiter and compressor made it pump too much no matter what settings i tried but still left those spikes in
in this case i would probably address the original drum recording. a good well mixed drum recording doesn't behave like that and doesn't need spikes cut off
@@APMastering I think i hit the snare little hard but somehow compression just increases the spike in the attack. SSL plugin. So i need saturation to round it of like tape would
it only works with tanh without post gain adjustment, so very specific. the reverser is not useful outside of a demonstration context but you can use the saturation sure
Hey, could you be nice enough to understand that some of us work in studios where we can't choose what DAW we're going to use cause they only have Pro Tools and please make AAX versions for Pro Tools? 😊
@@LeChapeauMusic you need to register and then deal with ilok and buy a dongle and sign with ilok and submit screen recording and then MAYBE they will allow you
@@puls3illegalmusic that would be good but they even made that difficult. even if i compile the plugin on my own computer it will be blocked in pro tools without all the ilok signing stuff
It certainly does, in about two seconds. Poor acoustics and a clinically deaf FOH-engineer can amount to permanent hearing loss. On the plus side, you don't need earplugs after the fact.
Is the saturation in your plugin similar to the method used by the Oxford inflator? I remember reading a description of how it works that sounded similar to
You can visualize the transfer curve by demoing Melda MWaveshaperMB and chosing the Mflator device. Then enter the edit mode to see the curve that resulted from your adjustment of the Curve control.
Does digital saturation actually work for this? Like are the algorithms actually gonna give you a decent sound when trying to tame peaks or is analog still the way to go?
@APMastering Roger, am I the only one that feels that digital saturation seems to end up sounding pretty much the same if not exactly the same even across multiple plugins Is there an algorithm that companies all use? Or do I just need new ears? 🤔
I think guitar amp sims prove that digital distortions can sound very distinct, and very good. That said, I do tend to prefer analog for subtle saturation.
@@Notinserviceij I'm in the opinion that it's all kinda samey in analog side as well. Tested a dozen different diodes, tubes and transformers by just clipping audio through them passively. At the end of the day they all just spit out even or odd harmonics (or a varying mix of both). Any noticeable difference is always due to something else, like the threshold where it starts clipping, the interference that gets picked up or the frequencies being filtered out. My ears are absolutely not convinced that an expensive AX7 sounds any different than a cheap AX7 with same gain rating and condition. When I'm swapping recordings going thru different transformers, it just feels like there's a tilt-EQ after a clipper. Diodes? Kinda interesting, because by combining different ones it's easier to mix different even/odd ratios. Super cheap also. Tape is kinda unique, but again not just because of saturation.
Your tanh function is a classic example of a soft clipper. Not all saturation has a limit. That is why hard and soft clipping are actually useful terms.
@@APMastering then what would you call polynomial functions that introduce related, signal-dependent harmonics but do not have a limit? F(x) = x + 0.25 * x^2, for example.
@@spicechateau i use quadratic polynomial functions in some of my (as of yet unreleased) plugins. it is 100% distortion but i'd not call it saturation as it bares zero relevance to analog saturation... what is it supposed to model? magnet flux sat? iron core sat? tube sat?
@@APMastering I think that’s a strange way to define saturation. Tanh is not a model of any analog saturator. That said, if we’re trying to stick to functions that roughly emulate physical principles, you can approximate a compressor’s ratio control with a piecewise function that adds saturation without a hard limit.
Clearly you’re doing a lot of plug-in work and have knowledge of the field, but I don’t think I’ve read a single paper or textbook that would agree with your definitions. It really seems like you just have a personal vocabulary and you’re calling people out for not adhering to it.
@@haslo_ The video is very long, so any comment that addresses it has to be reasonably long as well. Here's the short version of mine: "Good, you're advocating 'soft clipping', kind of like everyone else is, and just using a different term to describe it, which happens to be at odds with the rest of the way the rest of the audio industry has used that term for many decades now" 🙃
@@APMasteringThat's because you choose a very soft saturation curve. If you use something with a smaller knee, you'll certainly have information loss in the form of quantization.
@erinburke9711 I totally agree. I did not place enough emphasis on "given sufficient resolution". Given infinite resolution there'd be no loss,, but I agree that's not real world.
correction: THIS distortion can be undone great video as always mate! really informative! could you explain a bit WHAT clipping is and how clippers work and how it is and it is not just compression? i have trouble finding good info on that
there's probably tanh in ableton by i don't know what it's called, i don't use it. but if you have max msp in there you should be able to do it no problems!
good vid. except soft clipping is a valid term especially when an adjustable threshold is involved. yes it's just a form of waveshaping with an implicit clip threshold like you demo, however as a tool, an adjustable threshold implies you do want to hard clip above it in a specific place on purpose 🙂. so 'soft clipping' is a good way to think about it.
@@APMastering yes, any waveshaper can. but my point is that if the user is given an adjustable threshold, then clipping is the specific intention of that tool, rather than the byproduct. ie. the name is descriptive of the application, not the dsp.
@@desperateBeauty well i find that saturation is just this. i used to limit stuff by sending it through a tape machine for example. that's saturation by definition. nobody calls that tape soft clipping
wait I'm sorry but when it comes to the speaker issues, aren't they talking about physical clipping? It says "whether the mixer, amplifier, or any other piece of audio equipment clips the signal, Damage can occur without the amp's full output". If they are talking about physical gear causing the clipping i doubt that a loud mix that used smth like classic clipper will cause a pa system to go to shit lol
@@APMasteringinteresting i thought it would have something todo with the overloaded signal that is coming in instead of the waveform that it had to reproduce thanks for the info i‘ll look further into this!
@@APMastering isn’t constant voltage == dc? Ac is alternating power == audio signal. A hipass removes oscillation of ~0Hz which is dc offset which is constant power.
to be more specific, an IIR high pass will make the tops less flat if your Q is gentle because of the phase rotation caused but you achieve the same phase rotation with an all pass. if you use a FIR high pass, so just the filtering without phase impact, there is zero difference to the waveform of a square wave
@@APMastering ah so it’s because the number of samples clipped is low enough and alternating in sign that it is more like a momentary square wave rather than unipolar dc offset. I didn’t think about the basics apparently :D. So now we can conclude that a square wave will also break PAs. Synth bass players better change their waveform to saw then xD
This is again fake. Then take off distortion from AC/DC guitars please. You can't fkn redo clipping if its rendered. Unless you are RECORDING in 32bit floating. Also for the clipping drums thing... Clippping does make the drums cut through the song and its vastly different than just adding more volume with a fader. I don't know why you fear losing transients when you are using a daw which have transient designer plugins. With eq and a transient designers i can make even a bass guitar sound like an atomic snare. With limiters/compressors you going to lose transients on drums anyways so its your choice to add them back. Lots of times the drum tracks already have way too much transients and they just poke in too much in the mix.
Also, for what its worth - an allpass filter (or essentially any filter) will alter peak values when applied after limiting too, not just clipping. Perhaps if anything this is an argument to leave some headroom on the limiter ceiling
![Eminem - Fuel (feat. JID) [Official Lyric Video]](http://i.ytimg.com/vi/t5H_CewqpKA/mqdefault.jpg)
WINDOWS VERSION is now uploaded. PLEASE NOTE: if it wasn't clear from the video, my plugin does not magically undo ALL kinds of distortion. It is only a demonstration plug to show how, when you saturate something, you are preserving the details of the music despite the distortion and peak reduction, whereas when you clip, you simply delete your audio, loosing all information over the threshold. I think my plugin demonstrates this well but don't get carried away and think you can reverse all kinds of distortion with this, you can't. Feel free to play about with it but the reverser side really is a kind of trick for demo purposes only and I'm unsure you will get much genuine value from it. That said, the tanh saturation on the left of the plugin is very pleasant and useable, and if you can oversample it (like you can do in reaper for example) it will sound even sweeter..... ALL nonlinear processing is subject to some degree of aliasing and should be oversampled in the best case... 4x is fine but more than 8x is totally unnecessary (note to self: update the plugin with a second order antiderivative so it sounds great without OS....).
@@APMastering what? 😂
@@CFox.7 whatever
Thank you for the Windows version. I don't see it on the website, but I will send you my email so I can get it. Thank you!
❤
❤@@CamariMusic
This man came to disrupt the entire audio software industry 😎
I think the whole point of clipping is controlling the crest factor. It's not just loudness, but hitting the right compressors/limiters with a more uniform signal, so it's not like choking on a couple of transients and completely missing others.
that's fine but this isn't what i've been seeing people do online or my own mastering clients
@@APMastering it's a youtube tutorial problem or information availability problem, not quite a clipping problem itself, plus it also depends on a stylistic choice and genre itself. I love me some soft clipped or tanh kicks and snares.
@@APMastering so why do you say to don't use clippers, if you know that there's a valid reason for using clippers?
@@SenorTropiCat see part three, dropping in a couple of hours
But you can't undo distortion of existing audio that you don't know the 'transfer curve' of ....... which is the only practically desired use.
sure but the point was demonstration about information loss
just tried it on nasty distorted piano in stereo......thanks rescued a location recording...brilliant
really glad it worked! wouldn't have imagined it would have done.
@@APMasteringLol so many people here not understanding the point of the plug-in
@@erinburke9711 yeah but it's also fun to see people messing around and doing cool stuff with what is otherwise deemed basically unusable code ha ha
man I think I've been living under a rock.... people actually CLIP their masters????? I've been working for years in audio engineering and it sounds crazy to me people do this on purpose lol, as if squishing everything with a Waves L2 wasn't bad enough already.
Maybe in the Rock and Metal circles I usually work on it isn't part of the culture to clip things and its more of an electronic music thing, or maybe I just got lucky, still great video man! Blew my mind seeing things being de-saturated!
I was like WAIT A SECOND. but now I get it. Super cool video. I had no idea you could undistort something.
I think this is the sort of thing that Airwindows console plugins are doing, doing some sort of distortion on each console channel and undoing it as one on the master.
not sure but i think the code is open source so if anyone wanted to find it, it's technically public info
@@TheRenegadeScrew was thinking the same thing. Chris also talked about encoded and decoded signals too.
You should have a chat/collab with panorama mixing & mastering on the topic of clipping. You lads both very knowledgable on the topic and have opposite takes - would make for good content/discussion. Moreover, you both have a catalogue of work behind you so there is established experience in the industry behind the opinion.
yeah almost everything i watch from him triggers me which is why i should probably sit down with him and discuss lol
@@APMastering ive seen his stuff and i think you really do have a lot to talk about. to be honest, i used to have a very similar approach to his when it comes to getting rid of random peaks - i just clipped them. after watching your stuff which opened my eyes i got back to the traditional levelling approach.
@@LeChapeauMusic did it sound bad when you clipped the random peaks? Did you use just technical hard clipping?
@@SAFAMASTR the difference was barely audible. no matter how you do it, getting rid of random peaks makes the whole thing sound more... precise. that being said it's much more important when you're actually using a limiter in the end, something i almost never do. (or at least not in the way most engineers do it)
@@LeChapeauMusic i get this. Especially if you're mixing with high unrestricted headroom. But getting sound even close to the recommended streaming loudness of -13/-14 lufs usually requires some limiting. And the last thing you want to do is to go below, since they might boost and limit it for you, badly... And that's where clippers can be extremely helpful and actually get rid of noticeable artefacts (done usually by the limiter), not introduce them.
In the kick example - 100 samples is only 2 ms - it will be audible when you play a 100 sample/2 ms white noise sample in silence, but the same 2ms in a mix is not really audible at all - even less so when you include the fact that the initial transient of a kick is usually high frequency heavy, and is on beat with other instruments which helps mask the white noise.
I use clippers on drums and limiters on tonal instruments, and the tradeoff is worth it - the final limiter barely works more than 2-3 dbs before I get to stupid loud (-6 lufs or so)
The sad fact is clients always A/B their mixes and masters against other stuff they hear on Spotify. If its not similarly loud to other releases, then you get the blame. While I agree that there are some terrible examples of shoddy production, this should not reflect on the potential of the tools. Looking at badly deformed nail on the wall and blaming the hammer isnt the direction we should be taking.
this is not my experience as a mastering engineer
@@APMastering which part exactly isnt your experience?
Like the content and insights.
A key difference with tanh vs clipping is that with clipping
you can affect a transient with no change to samples below
the threshold. But with tanh you must accept the gradual onset
that will modify samples below the threshold. Both result in a
distorted signal, and which one is more noticeable is program and
threshold dependent. Assuming undo-ability is not desired
then the choice comes down to what your sound goals are.
i agree but the nonlinearity of tanh doesn't really bother me until the signal is real hot. pretty sure i could run a string concerto through tanh and nobody would notice given a bit of headroom
i start to like this guy, ruffeling a bit of feathers. he got some goods points tho
There's a deliciously unsavory irony to be found in listening to a comparison of clipped vs non-clipped signals through the RUclips smoosh-filter. Honestly, I could hear no difference in that snare example. BUT I 100% agree with your conclusions.
@@periurban were you listening at 1x speed? feel free to listen to speech at 2x but if you don't listen on 1x to music yt eats the transients
Yeah I couldn’t hear a difference either.
@@APMastering Listen at 2x? I tried it once and it made my soul hurt. No, I listen at 1x always. BUT I am in my seventh decade now, so if the juice is flowing above 10kHz it flows without me.
@@katielowen Do you have good hearing? Mine is fading fast!
@@periurban I think so, and I listened to it in a car!🤣
A demonstration of the invertibility of continuous strictly monotonic functions.
It's not just about loudness, it's about not loosing punch while making it loud.
And the signal is going to be relatively louder than other songs and will be percieved as better.
It will even sound louder when turned down by a streaming platform.
Even phase rotated, it will still be louder.
There is nothing a listener hates more than having to readjust the volume for every song.
Increasing the volume of a quiet song in order to match loudness of the soft-clipped song is going to cause amplifier clipping.
And distortion can be undone no more than any process can be undone with an undo button - if you have the original information or the information about the process applied to it. You can also undo clipping if you lower the volume so it's not clipping anymore.
"And the signal is going to be relatively louder than other songs and will be percieved as better. "
No it won't be. If it is loud but sucks, it will be deemed a bad mix and the listener will turn it down. Average listeners might not be able to talk anything technical but they know a bad mix when they hear it.
"It will even sound louder when turned down by a streaming platform."
This is completely false.
"There is nothing a listener hates more than having to readjust the volume for every song. "
Do you have evidence?
"Increasing the volume of a quiet song in order to match loudnes of the soft-clipped song is going to cause amplifier clipping."
This is wrong. Speakers and amps can typically deliver double the peak power as sustained power. A clipped song decreases the crest factor, meaning the song has less peaks and much hotter sustained power with less loss through kinetic energy which puts more strain on the amp and can overheat the voice coils.
"You can also undo clipping if you lower the volume so it's not clipping anymore."
No you can't if it is committed. The point in this video is that saturation maintains detail AFTER BOUNCING. Whereas clipping destroys it.
Hope you can learn something from my response.
@@APMastering You have an idealistic view of the world. No, average listener can't tell a fart from bass guitar, let alone recognize a good mix.
Loud songs do not suck by default, this a totaly misplaced argument. Louder is percieved as better even by many mix engineers, you bett an average listener is going to percieve it as better sounding.
"This is completely false" - Try streaming normalized songs, there will be difference in percieved loudness.
"Do you have evidence?" - Funny how I have to prove things in comments, yet you don't prove your claims in your video.
I and most people I know, hate readjusting volume, and if anecdote is not enough for you, it costs energy and it's a distraction to have to go to your amp, or look for your phone or the remote, when you just want to enjoy music, even worse, if you're busy doing something else while listening like most people do.
In practice, speakers are typically not going to be fed unaltered squarewave unless the clipping is occuring internally.
Your arguments are really rationalisations of your biases.
A commited song can be replaced with the unclipped version of the song. Your point is no different, you need to store information either about the original signal, or about the process that's been applied to it.
I hope you apply logics and realize that you aren't teaching, you're just expressing your point of view.
@@duncan.o-vicyes the average listener can tell a fart from a bass guitar
@@GoldenPickaxe It was an expression, a figure of speech, most people overestimate their hearing abilities.
@@duncan.o-vic I would say, that a average listener can say, if a song is mixed well or not. That for sure.
Loud songs, which are pushed need to be very well EQ´d. There are a lot of things which go into having a great sounding, loud mix. Starting with the low end.
But one point I want to say as well. I sure as hell hate, when I need to fix my volume while listening to a playlist lol
But I also do listen without loudness normalization, which one CAN definitely turn on. This will help to avoid that, but you also then not hear the mix, as it was intended.
I knew. :) Distortion is, of course, a "linear" process (minus the sample/bit resolution reduction). I used to do this years ago with an ancient distortion plugin where I could make a custom curve. I could 'magically' clean up old recordings to a certain extent, except clipped audio. I thought most sound engineers knew this, though!
not sure what you mean with linear but undoing distortion is pretty esoteric afaik, cool that you were doing that ages ago, i never was so successful at it in restoration work
@@APMastering What I mean is, it can be undone. It doesn't introduce time-based distortion like compressors. i.e. You cannot expand a compressed signal back to the exact original, but you can undo waveform distortion if you know the curve used to distort the waveform (without clipping, of course).
EDIT: I know "linear" probably isn't the correct term to use when it comes to waveform distortion. Can't think of the correct term to use.
theoretically you can reverse compression but it's just impracticality hard to do. i tried it for my compressor series and wanted to pull my hair out
@@PrincipalAudio The Dolby systems... compression, then compression undone.
This is an excellent approach to showcase the reverse function to a broad audience. As I can see, it is not applicable for 32-bit (float) signals, as we lose precision with each iteration, but the signal samples at 16-bit/24-bit depth are considered integers, and everything should work just fine. Anyway, I still suspect there can be corner cases.
I like your channel!
This is the kind of stuff that every real Engineer should know. Thanks for all you have done for us and you help us out tremendously. Great videos explaining everything.
thanks!
Can this plug-in reverse distortion on a file where the distortion was not first added by this plug-in?
only if it also used a tanh to 0db algo, otherwise the mapping won't match up
@@APMastering So basically one needs to know which algorithm was used to undo it.
@@gulagwarlord yeah. AND the ceiling
@@APMasteringis there any reliable way to learn this information from a track? I have been told in the last Waves makes plugins that can undo brickwall limiting and compression and restore dynamic range to modern releases
So, useless.
So, a certain TYPE of actual distortion (in this case algo of hyperbolic tangent, which compresses intensely the original values of the sound into a hypo-tan set of numbers). This can be undone by the inverse function. Makes sense. Clip is dangerous because the original harmonic content has been wiped out and no longer available.
It may make sense to saturate gently, but having clipping usually is going to be quite regrettable, especially if it is printed permanently into original multi-track sound. I'd agree that "saturation" is not really clipping per se. It has similarities but it's not really the same.
Hence why music production is not just objective it is also subjective hence why the tools are used and abused for creative uses not just for the technical aspects. Best of both worlds.
There are interesting uses of tools and then there is digital information loss
@@APMasteringbut digital information loss can be a desirable too. See bitcrushing, samplerate reduction, deliberate aliasing, low bandwidth codec plugins for things people use
This is kind of operating under the presumption that clipping is the only process contributing to loudness for some people. Typically clipping with K-clip is just the final process after saturation, compression etc to control those 'every now and then' unwanted transient peaks to prevent higher peaks in up stream bussing at least for me. K-Clip might not even meet the strict definition of clipping you've laid out, so saturation is really what's putting the work in loudness wise for me. I never want the clipping to be audible or adding any kind of unintended colouration, I just want mixes with the kind of marriage of punch and density expected in a given genre.
I agree that mixing loud is MUCH more than clipping. In fact you can mix loud completely without clipping. I think kclip is incorrectly named. It's a multi band distortion plugin.
I think you may be triggering youtubers who rely on the esoteric way to describe plugins. I appreciate it though. the producer crowd sometimes feels like they use vague language on purpose to make themselves sound more skilled. Whereas the music performance youtubers seem to be the opposite. Play this scale in this way with ghost notes to get a Swing saxophone effect. Versus "use my effects chain to make dope beats". It's infuriating, honestly.
maybe but i think there's this other dude with 30 extremely long videos on clipping that built a following and people think they need to use 100 clippers in their mixes
@@APMasteringit’s 102 to be fair
@@APMastering is he super hardcore? And may or may not own a plug-in company 😂?
@saxmanash Selling sample packs is something I can still understand, selling effect rack presets - sure, why not.
But nowadays these people are also selling midi chord packs, and apparently getting sales. This is something I will never understand, with a bit of reading into music theory you can construct any chord you want, why are people buying digital note values… Are these the same people that buy NFTs? 😛
@@Wizz15 yeah i don't get that either
thank you for saying so much truth about these topics where so many wrong advices are said in the music production community!
This is an excellent video! I’m so glad you’re pushing this theory, so many tracks are so heavily clipped nowadays.
Great takes. Like this channel very much. So what are clippers good for?
almost nothing. i genuinely can't think of a single valid use case in music apart from maybe synthesising a square wave from a sine wave but ironically it's more difficult to create a sine source than a square source anyway
i'm definitely also more a fan of softclipping than hardclipping but i wouldn't go as far as to say that hardclipping is never useful and just makes things worse. the harder you clip something the more direct and close it sounds. you also have an easier time controlling the peaks and get stronger harmonics if you clip harder. but there are a lot of sounds that really don't like being clipped too hard, like vocals. clipping is also useful before compression to make the compressor react more evenly to a signal, and it's useful after compression to get an even sounding distortion, but that is true for both soft- and hardclipping
I think we have slightly different approaches... maybe im more old school but I never use clipping. I feel like there are many other processes which I am happier with than clipping that achieve similar processes, or I find the processes unnecessary.
You videos offer very high value. I have always been dubious of clipping. And came to that conclusion after rigging up a couple of anti-parallel germanium diodes in the signal path of my mastering chain many years ago. Indeed the only audio that should be subjected to such shenanigans is the signal from an electric guitar...! OK...perhaps some synth stuff as well... ;) Just not mix busses or masters!
Unpopular opinion: If it sounds good it sounds good. If those bits sounded bad then delete them. Genre dependet i think resloution is overrated. I dont care how the effects are done, as long as it serves the music. Sometimes there is a use case for hard clipping before a process like compression regardless if there is a phase rotation after. Maybe soft clipping is the move, maybe not. Ears are the final judge.
ears can be deceived by loudness. always level match your processing, especially something as destructive as clipping, like I do in this video
Oh I do, but sometimes the data loss is worth it
There was a "clip restoration" function in the old Adobe Audition - I had some success with it on badly recorded / distorted audio. Didn't suit everything, but when it worked, it worked very well, and saved somebody's session.
yeah this is not really getting the original back though, it's interpolation
sweet, good work on this one! looking forward to a version that will possibly allow this without it being originally added to the song. 🙌🏾
Hey nice video mate! I don't see however why you treat information loss as a bad thing.
While mathematically you're right, saturation does reduce the volume of the information, and nobody does atanh functions on their music.
Clipping just lowers it to zero. There isn't actually that much of a difference, is there?
About the PA systems, I guess that means you can't play square waves either? I don't find "damages the PA system" a convincing argument unless you're mixing specifically for speaker longevity. It's a side effect.
And the argument about clipping mixes applies to sat as well.
This reads to me as a bit dismissive of clipping in general? I think if it sounds good it is good, and while I normally think saturation does sound better, there's a place for hard clipping.
Specifically, one advantage it has is that the parts of the audio that aren't clipped aren't also altered in any way - they're just turned up. This is very useful if you just want to turn down a peak.
a loud square wave can kill a pa
People use clipping and distortion for different reasons, distortion is changing the sonic character, clipping cuts off the inaudible peaks, so they won’t trigger your bus compressor. These are not equal steps in the prodction process, both have their valid place in the workflow.
And why anyone would want to reverse the distortion made by a specific plugin? In case you’ve lost the project file? Otherwise you just go back and export it with less distortion…
the reverser is just a demonstration. its not intended to be used restoratively
@@APMastering okay, and the other half? That saturation is an artistic choice, while clipping is a utilitary thing, that helps your bus compressor’s work? I think you’re forcing this idea to fit in your concept about “clippers are a scam”, but it doesn’t hold up. Cutting off inaudible and unimportant transients on a drum hit is a real thing
@@SenorTropiCatbruh if you do this with "free" plugins then a 200 dollar clipper is a scam
I don't care what creative purpose you give something
That's just because you are tryna cope with spending 200 bucks on a piece of software that is not worth that amount of money
Regardless, stop spending money on plugin they don't make you better
@@SenorTropiCat I think clipping can be creative too, but I'm into harsh noise and other extreme genres as much as I'm into more conventional styles. I feel like there's a dangerous dogma around the idea of resolution and quality that -- although it might be more based in data -- is ***similar*** to how audiophiles think of audio. If it works for the genre it works. If hard clipping a snare sounds better than soft clipping a snare I'll happily delete those bits and never look back. One of the most useful things I learned in my undergrad studies was that there's no such thing as a bad sound, just a sound that's wrong for the context. I feel like those nuances get lost in these discussions.
@@Notinserviceij who said i’ve spent 200 dollars on a clipper? 😹 I’m using the clipper in the TDR Labs limiter, a plugin that I would use otherwise too… Even without the clipper part.
Hi 👋 😂 I was “that guy” in the last video. You’re not necessarily wrong here (nice to hear someone talk about the reverse saturation option. I discovered this when playing with waveshapers thought it would be a cool video idea. I like to put processing In-between for weird sound design experiments… anyway) there’s a couple of things I disagree with again.
Hard clipping is actually cleaner than Saturation. Why? Because your signal is spending less time in a “non-linear environment” Intermodulation can be reduced using hard clipping. At least it sounds cleaner on complex waveforms like a mix.
100 samples out of 44100 per second is still very little. The question is do you hear it? If the answer is yes, you’re doing it wrong.
I do often use “softclipping” as it is “softer/less instant” but at the end of the day you could argue this is also reversible. But it really isn’t because of the signal is loud enough we’re limited by our bit depth to realistically restore that almost.almost.almost clipped information. and realistically the only difference between hard clipping and saturation is how steep is the non-linearity. (Assuming symmetrical +&-)
Also aren’t all speakers high passed and low passed anyway? Meaning any “prolonged energy” would be offset by the phase shift in the speaker itself? This destroying speakers argument is more to do with sheer volume than the shape of the waveform.
…. God this is a deep topic. Appreciate seeing a different point of view but I still think clipping and “soft clipping” are very useful tools.
Also no shade here. I appreciate the video. Nice to nerd out with fellow nerds.
Hard clipping can be better as fewer values in the waveform are being modified as well.
@@dodgingrain3695 yes that’s the essence of what I’m saying about having less perceivable intermodulation with a clipper compared to a saturator. Meaning it can unintuitively sound cleaner. Clippers get a bad rep but they’re actually the most transparent peak control tool out there. When used properly.
The whole cant undo information loss thing makes me want you to base part 3 around AI tools meant to restore information. sony DSEE HX, Izotope Spectral Recovery, voice fixer, heck even the old low bitrate restoration trick of cooying the track shifting it up an octave and EQing everything below the cutoff on the real thing away
Izotope declipper i have had make some good results Audacity not so much but maybe its for more specific issues
although I'm a mastering engineer I dont tend to do much actual restoration work, so this isnt really my strongest area. However, although I have used izotope rx in the past to sort of improve some really terrible stuff, I don't think any of that stuff is a substitute to getting it right in the recording, as none of it really sounds that good. I would be interested to interview someone who does actually restore historic recordings etc but this is not my thing at all.
Am i lying to you?
Yes you are.
@@Durkhead Save yourself two letters and a moment of your life - "broadcast" is the correct word, it doesn't need the "ed."
i'm not wrong about all pass on radio, look it up. i'm also not wrong about speakers, if i were all of those reliable sources would be wrong too obviously
So in terms of damaging speaker cones and what not, essentially the idea is that hard-clipping introduces DC offset to the signal (which basically means that the signal has more positive or negative values overall).
But doesn’t heavy saturation also introduce some amount of DC offset? Especially when applied to non-symmetrical signals?
So, I don’t think the kind of distortion really matters in terms of DC offset. If the PA system is handling audio properly, and the audio signal is being High Passed to cut out the DC, either in the actual recording, board, or in the drivers, then the DC offset thing is kind of a non-existent issue. Audio signals aren’t even truely “flattened” when being clipped anyway, because of the nyquist limit. There’s still some actual wiggle in the signal (recall: true peak)
So correct me if I’m wrong, because I’m mostly going off of intuition, but I believe the issue only really arises when the DRIVERS themselves are being clipped. THAT’s where the issue comes into play. And that’s not an issue of the audio itself, but rather an issue of the input volume simply being too loud going into the speaker. But if the drivers aren’t being clipped, an audio signal with a clipper on it isn’t truely “clipping” bc of the Nyquist limit (hard LP filter). And any dangerous amount of DC offset would be mitigated with a HP filter.
sorry for short reply but has nothing to do with dc offset in the normal sense, unless by dc offer you mean the clipper top and then that's not really the correct term. it's overheating because of power without movement
@@APMastering no worries. My point is that “clipping” on a sample level is not actually power without movement, because of true peak: the samples don’t reflect the actual physical “movement” of the speaker cones, but are something of a “guide” for the signal.
Also, DC offset is the term for power without movement (theoretically, a 0hz or near 0hz signal)
sure but look at my low pass filler example in this vid. it's almost the same
@@zionjaymes4415 of course but you don't say dc offset for part of a signal which is demonstrably above 20hz
Magic trick keeps original undistorited signal ready waiting for that undo distortion function to be activated.
false. download and try for yourself by first bouncing out the audio
This might be missing thing from digital audio i've been looking for you got here. I've had a problem with getting drum compression sound the way i want. I always seem to not be able to get that smoother fat tone and end up with eather too much spiky transient or pumping. Drums sounds i like are from the analog days and tape compression is just like that, it rounds the sound. I think the flat waveform tops are what makes also human ears tired when listening some loudnesswar stuff with headphones you feel this constant push against your ears but if you listen to something older that has tape compression and tupe saturation in it, it can be loud but will still sound smooth and more pleasing to the ears. If your plugin is a way to do a digital version of tapecompression like it seems to be it's really needed. If you have it also really clean without aliasing or similar wierd harmonics (i wrote this in the half point of the video)
try parallel limiting or distortion
Glad to have you explain all this and that basic digital facts really helpful to know, good stuff!
does that mean if other plugins had this opposite algo, they would also be able to undo ?
any strange cool sounds emerge by putting other effx between the sat & the reverse ?
yes.... REALLY nasty distortion ha ha, you can try it. it's pretty impressively glitchy
More click bait, just sad, and sad that people believe this guy. Again, I don't think you understand the use case for a clipper. Obviously a clipper tends to make a drum hit feel like it has less punch but it does make it louder because you're TURNING UP THE GAIN. You could maybe argue for a more feeble sound but it won't be quieter. Nobody clips a signal and then turns it down either, ROFL. Also referencing Ethan isn't helping your case. With speakers I think your also confusing a square wave with clipping, they are not the same. Square waves are used all the time in music especially as a bass sound and it doesn't inherently destroy tweeters. Speaker manufacturers are not worried about squared off waves, they are worried about excessive high voltage. Case in point at 11:35 "its sustained power", not squared off waves. The "unknown reference point" is NOT the users volume knob, its the previous song. Distortion, regardless of the type and regardless of the chain in lufs is perceived as louder even if it isn't, clipping is just another distortion. Seems like your best argument is that someone that doesn't know how to mix well won't be able to get a mix as loud as someone that does know how to mix..... which has nothing to do with clippers. So much other stuff is wrong in this video, there just isn't time to comment on it all, but hey one last one, the notes from the book at 18:05 are essentially telling you to use hard clipping and avoid any kind of transfer function that isn't hard clipping. Also that book is taking the standpoint of clipping only from a utilitarian view and not as a potential creative tool.
so much wrong here i don't know where to start
Well done AP - you're kicking it out of the park!
thanks mate
Ive been waiting for this to become a thing, since the latest wave of truly brain-erasing AI assisted noise reduction and de-reverb plugins started surfacing. Let's say,you've got a spanish guitar thats badly clipping. If If a human can visualize what it sounds like without the distortion,it should be possible to use generative AI to pretty faithfully reproduce a clip without the distortion.Amazing time in DSP processing for audio.
yes! but the way I see it, part of the greatness of great recordings was the time, place, coincidental stuff that all came together. AI will make everything great and then relatively speaking nothing will be great any more.
@@APMastering Dont disagree, but that is neither here nor there, not unlike how an AI generated vocal stem will never be 100% exactly like the source. Damn close,yes, perfect, no. The new paradigm, that if a human mind can visualize what an undistorted verson of problem audio would sound like, than a neural network can, too, and so, this should be possible. (popcorn!) haha
@@APMastering If you would have asked me 5 years ago if this would be possible one day soon,Id have to say no way. Imagine that just a few years ago it really looked like we were at the limit of what could be done and how of how an audio signal could be manipulated digitally or otherwise. Im looking forward to what this means for restoring old damaged recordings, forensic audio, moreso than music,per se. In the future, i believe music producers will no longer play musical instruments, nor have or have to have any technical skills moreso than any other person. if you want to make a song, just think it. Like most aspects of ai, this both fasinates, and saddens/terrifies me. Someone wake me up when ai does something tho help, not ruin society at large. :) Still though, may you live in interesting times!
sorry, but after watching this video all I think is...
my dude got gaslit into believing that we clip entire master in electronic music or hip-hop or smth with +15db clipper or smth... or we clip the everliving crap out of drums, information loss, blah blah... great stuff.
again, skill issue from rock musicians, mixing and mastering engineers trying to chase the high from high lufs.
dudes just got sent 20 years ago back into loudness wars insecurities. so sorry to people that have to go through this.
@@yeshello2528 troll
@@APMastering le epic trollge moment
@@APMastering In what way?
Just a question
The saturation in your plugin is a achieved by a linear process. analog saturators are based upon nonlinear processes that cannot be described by the means of transfer functions. In that case, it is not possible to reverse it. Am I right?
no, tanh is NOT a linear function, its a nonlinear function.
A linear process is something which can be described as output = SomeFunction(input), as in this case. A nonlinear one can't... However, I don't know how nonlinear are the equations of saturation processes inside real hardware. In my opinion this video should be called "plugin saturators can be reversed". Good video btw
@@veinsbandofficial6221 again i dont agree with what you say linear means, it's a well defined term. But sure, hardware likely has more complex distortion characteristics than digital tanh
@@veinsbandofficial6221The word you're looking for is time-invariant
@@veinsbandofficial6221Or static
Dude, you’ve started a revolution against audio information Bullshit
🍻 cheers !!!!
Actually he is just presenting strawman arguments in order to fabricate controversy.
@@danyavilaoficial “clipping is bad” is the bullshit.
It’s an outdated narrow minded dogmatic philosophy. Any tool can be used for harm and good and the audio engineering community should embrace clipping as the useful tool it is imo.
@@duncan.o-vic Don't forget the forthcoming "Pro" line of his plug-ins...
Hard clipping can potentially create a DC offset because the straight line of a square wave is DC.
A DC offset can potentially mess up your amp and speaker biasing, causing damage, heat, or extra distortion.
However, this is only theoretical, as all of them have so-called coupling capacitors that block DC.
So, I don't think that square waves will damage your speakers on their own, it's common practice to use square waves to calibrate amps, speakers, tape recorders, and so on.
The dangerous clipping they are referring to occurs when you feed your signal to the speaker or amp so hot that it exceeds the power rating of the system, overly loud source material only adds to that.
this is wrong. Although the top of a square wave is static, it's not electronically considered DC since it is not 0Hz... let's say the square wave is 1kHz. this is higher than 0hz therefore it's not DC. being reductionist about just the top portion is not how people normally discuss this
@@APMastering are you sure that waveform analysis in digital is correct? god knows what might happen during dac conversion...
@@APMasteringTo achieve the flat line of a square wave, there must be a DC component present. Even if the wave is 1 kHz or higher, it still contains that component. You can test this yourself: generate a 1 kHz square wave and apply a 1 Hz high-pass filter on top. Then, observe the oscilloscope to see how the wave changes. The flat line will drop even if the filter never affected the fundamental frequency, because it removed the DC component.
@@eliashowe7115 no that's not what DC is. DC is 0hz. high pass filter at 20hz and see what happens
@@eliashowe7115This is not right. Clipping can essentially induce DC, but not because of the flat line. It's because the energy imbalance in the top/bottom swing of the waveform becomes exhaggerated, depending on the material. You can usually see some DC on an analyzer after clipping.
Great video, it's refreshing and rare to see people exploring how things actually work and providing a spring board for constructive discussion! Does the same information loss apply to hard knee limiting where it seems to flatten a peak instantaneously before having a gradual release? Is this the same as a hard cliper on the onset while the gradual release is mitigate some distortion for cutting into the waveform or is it somehow compressing the dynamic range instanteously in a different way? Does that mean a soft knee limiter is functioning the same as a tanh function on its onset and by this measure would you only ever personally use a soft knee limiter?
i need to study the dsp behind limiters more. i going to try and build one soon. up until now i've just used them
@@APMastering Keep us updated on your findings, it would be silly to be avoiding information loss with 1 process only to be unknowingly doing it with another. I'd love to see a video on the process of building these plugins, including the research/circuit analysis, coding and UI/UX if you find the time.
I was searching for SOOO long for a device like this! ❤️
don't get too excited though, this will not magically remove all distortion. It SPECIFICALLY remaps tanh algorithm saturation
Can you do a video on newfangled saturate. Its doing something different and I'd love to know the downsides vecause it sound freakin awsome and loud but it kills the bass a bit. Its a clipper so the name is kind of misleading
website is not working
i hard switched to new servers last night, should work now! let me know if it's still down your side
Tanh is too soft for mastering use unless you're happy to distort low level samples, which generally isn't desirable across a whole mix. The term "soft-clipping" is generally used for saturation curves that are much harder. Closer to full on clipping, while retaining a soft knee. For example: x/(1+x^8)^(1/8)
sure my free plug wasn't aimed at mastering engineers. As a mastering engineer I don't typically use this kind of processing anyway. Having a steeper function has its uses maybe as a part of a limiter design but I don't see how it is so useful in mixing personally.... I guess I'm just not cool enough to want to nuke the dynamics of my music unlike many of the people who commented on my last video. Anyway I tried out your math....
channelData[channel][sample] = channelData[channel][sample] / pow(1 + pow(channelData[channel][sample], 8), 1.0 / 8);
imgur.com/a/1r0XHZs
@@APMastering K. I just want people to know they don't have to jump all the way to tanh if they want to avoid hard-clipping. For me hard-clipping is full avoid, but a slightly soft knee does have its uses, especially on drums. Once you get to tanh I would call that saturation, not soft-clipping, but all are part of the same spectrum, with only hard-clipping being strictly defined.
ok so your point is that soft clipping and saturation exist as separate terms, based on knee, and although there's overlap you are willing to say that soft clipping exists in between clipping, and that saturation definitionally doesnt get close enough, therefore the term soft clipping is required.
What I'm saying is, saturation is everything of this sort which is not clipping, and there is no room for the third term, soft clipping.
We understand each other. But we disagree. I will stick to my guns as I'm not convinced that saturation is inconsistent with what you call "soft clipping". I'm open to having my mind changed but I see no reason right now to change it based on mere popularity of the term. I'm sure I could demonstrate electronically a type of saturation which comes close to clipping, like for example saturation of an op amp based amplifier which reaches the power rails.
Question: I noticed that if I run a mix through a dithering plugin like the stock dither that comes with Cubase, using 16bit dither, I can drive the signal, it appears to "clip" (reaching 0db) but with no significant artifacts in the signal being introduced. What could be happening here? Is a dithering plugin acting as a "clipper" somehow?
I dont use this of course - I was just curious after messing about with various dithering plugins.
it might have a built in clipping function. im not familiar with this specific plugin since I dont use cubase (well I haven't done for around 20 years).
@@APMastering Thanks so much for replying! And thank you for such instructive videos!
you're getting into airwindows territory now lol
because i'm releasing a bunch of free plugins?
@@APMastering If I understand them correctly, Airwindows' Consoles work in this way. Saturating at the channel stage and de-saturating at the bus stage. The de-saturation isn't a perfect null because of how the tracks sum and play together, giving us this sense of "glue." Feels like things are saturating and working together instead of individually saturating and still playing back separately. Chris would obviously be able to give you a better explanation though!
i doubt it's the same because if you mismatch the saturator and reverser even a tiny bit you get crackle doom
@@APMastering and now I think about it, you and Chris should consider collaborating. You two have similar messages in your videos and products. You giving his plugins some much needed love in the GUI side of things (and possibly AAX support) and combining your processing philosophies would be killer. APWindows could change the game
@@APMasteringhow bad does it sound with a tad bit of gain-matched EQ before the de-saturation stage?
would you debate your viewers on this topic? I feel like, again, this is quite an interesting topic because of a lot of information missing from the perspective of both hip-hop and electronic music. for example, there is only a handful of people that clip individual elements like that. that one baphometrix video doesn't actually show how it actually works.
how would the debate work?
@@APMastering maybe also add live stream for audience factor and there you go, more content to learn from
Are you planning to expand this plugin to have more trransfer and... untransfer (ig just mathematically inverse)... curves? Seems really cool. I am curious what your take on De-Clippers is, like Izotope or Adobe or FL Studio. Obviously the original it can't be recovered, however the ability to "redraw" the wave definitely seems even more interesting after seeing how your plugin works in this vid
de clipping is impossible. you can only interpolate. if you want more drive for the saturation, just boost the level up before the plugin. you can boost 1000db and the output will still not exceed 0db 😎
@@APMastering I meant outside of the saturation context - more like for dialogue and stuff. I guess the "interpolate" you mention is the "redrawing the wave" with tools like the RX, which is what I was curious about your take on. I understand de-clipping in the literal sense is impossible, but what I'm more so trying to ask is about your opinions in regard to tools like the RX De-clip, Adobe's De-clip, the FL Studio De-clip is - outside the context of mixing, saturation, and distortion - for things like when a voice actor might get too loud on a certain word or line and clip the analog input into a recording, etc.
Trust me, I know there's limits to tools like that, but I'm interested on your feelings about tools like that and your insight into how they might work due to your level of in depth explanations.
@@5adb0i i don't use them because if someone sends me a clipped mix, i simply reject it.
@@APMastering I understand that completely. I guess what I'm asking about is not really the kind of stuff you work with on a day-to-day basis.
interesting concept and plugin. my question is though, does the plugin only reverse the type of saturation thats also produced by this same plugin or will it effectively eliminate other sat types/shapes?
only tanh
you need the inverse of the original distortion function
@@APMastering Thanks for the reply homie
Also, one of the electro IDM gig at the local club melted the PA 😆. That shit is real.
ha ha
i think that's just idm issue, they just do not give a fuck.
Do you think oversampling is necessary for distortion and extreme compression?
sometimes it can sound better
Another very simple and useful tool and a super educational video, fantastic work! Just got the windows download. I appreciate all your free creations and while I'm not supporting you financially, I'm happy to support this content by word of mouth.
How are you doing mentally? I'm imagining there's lot of brand new stress and anxiety that you're exposed to lately, with every vid being a controversy of some sort lol. Take some time off from responding to hate comments once in a while, it'll screw you up eventually and you'll end up responding to everyone with same energy. No good! So stay humble and healthy! :D
thanks! yeah Ive got a lot on right now but I don't really let the haters get to me!
I will try your plugin on my song where i don't like the spikes of the drums attack that sounded bad with limiter and compressor made it pump too much no matter what settings i tried but still left those spikes in
in this case i would probably address the original drum recording. a good well mixed drum recording doesn't behave like that and doesn't need spikes cut off
@@APMastering I think i hit the snare little hard but somehow compression just increases the spike in the attack. SSL plugin. So i need saturation to round it of like tape would
@@sinenkaari5477 make the attack longer and reduce the amount of gain reduction
Cool stuff man....the world needs channels like this.
thanks!
Agree... this is life-saving
@@luisaguilera6986it's really not, he's knowingly spreading false information to get clicks on his videos
Interesting! Can you demonstrate this using other plugins that are adding distortion or does it only work with the distortion added from your plugin?
it only works with tanh without post gain adjustment, so very specific. the reverser is not useful outside of a demonstration context but you can use the saturation sure
Hey, could you be nice enough to understand that some of us work in studios where we can't choose what DAW we're going to use cause they only have Pro Tools and please make AAX versions for Pro Tools? 😊
i feel you but avid gatekeep their format and releasing in aax is about as pain free as smashing you head against a brick wall
@@APMastering really? is it true that you need a license and stuff to release AAX plugins?
@@LeChapeauMusic you need to register and then deal with ilok and buy a dongle and sign with ilok and submit screen recording and then MAYBE they will allow you
Bring a flash drive with reaper to the studio and install it when they aren’t looking 🤙🏻
@@puls3illegalmusic that would be good but they even made that difficult. even if i compile the plugin on my own computer it will be blocked in pro tools without all the ilok signing stuff
I wonder if clipped content also wears out ears in a similar way it does drivers? Another episode?
lol it wears out mine as a mastering engineer
It certainly does, in about two seconds. Poor acoustics and a clinically deaf FOH-engineer can amount to permanent hearing loss. On the plus side, you don't need earplugs after the fact.
i wonder why did noone else come up with this??
its not that useful because the reverser is just remapping tanh and if it wasn't distorted with tanh, it won't reverse
This was really cool, would love it if you ported it to PC
coming tonight
Undoing digital clipping is impossible, because of loss of too much sample/sound information.
On everything audio info is gone there is no getting it back. Unlike energy in the universe
@@morbidmanmusic Yep!
Is it good for ''DI-ing'' an amp recorded guitar for trying out amp sims instead?
is tanh good for amp sim? not sure, try it out! 😁
@@APMastering I meant your plugin... I don't know much about math or tanh. I will try it out. Thanks :)
Is the saturation in your plugin similar to the method used by the Oxford inflator? I remember reading a description of how it works that sounded similar to
might be but I dont know what they used in there
Yeah basically. Inflator is two functions similar to tanh, with a "curve" control crossfading between the two.
You can visualize the transfer curve by demoing Melda MWaveshaperMB and chosing the Mflator device. Then enter the edit mode to see the curve that resulted from your adjustment of the Curve control.
Does digital saturation actually work for this? Like are the algorithms actually gonna give you a decent sound when trying to tame peaks or is analog still the way to go?
you can get a good sound in digital these days
@APMastering Roger, am I the only one that feels that digital saturation seems to end up sounding pretty much the same if not exactly the same even across multiple plugins
Is there an algorithm that companies all use?
Or do I just need new ears? 🤔
I think guitar amp sims prove that digital distortions can sound very distinct, and very good. That said, I do tend to prefer analog for subtle saturation.
@@Notinserviceij I'm in the opinion that it's all kinda samey in analog side as well. Tested a dozen different diodes, tubes and transformers by just clipping audio through them passively. At the end of the day they all just spit out even or odd harmonics (or a varying mix of both). Any noticeable difference is always due to something else, like the threshold where it starts clipping, the interference that gets picked up or the frequencies being filtered out.
My ears are absolutely not convinced that an expensive AX7 sounds any different than a cheap AX7 with same gain rating and condition.
When I'm swapping recordings going thru different transformers, it just feels like there's a tilt-EQ after a clipper.
Diodes? Kinda interesting, because by combining different ones it's easier to mix different even/odd ratios. Super cheap also.
Tape is kinda unique, but again not just because of saturation.
this is great but we need a windows version
i'm on it right now
Awesome! So saturation rounds of the peeks and this produces the extra harmonics that distortion is?
more or less
Big ups to you. Very educational
I does not work on mac, doesn't pass sonoma security...
that's not an error. its not signed. you need to bypass it.
@@APMastering I know and I did that multiple times including restarts. It still does not show up in Logic or LUNA.
Are your plugins Mac only? No Windows?
i'm compiling it now
No windows plugin available ?
it's up now
Amazing.... but you dont have this for windows?
just need to compile it. will do tonight
@@APMastering amazing.. thank you so much
Your tanh function is a classic example of a soft clipper. Not all saturation has a limit. That is why hard and soft clipping are actually useful terms.
all saturation has a limit otherwise it has not yet reached saturation. that is the meaning of the word saturation.
@@APMastering then what would you call polynomial functions that introduce related, signal-dependent harmonics but do not have a limit? F(x) = x + 0.25 * x^2, for example.
@@spicechateau i use quadratic polynomial functions in some of my (as of yet unreleased) plugins. it is 100% distortion but i'd not call it saturation as it bares zero relevance to analog saturation... what is it supposed to model? magnet flux sat? iron core sat? tube sat?
@@APMastering I think that’s a strange way to define saturation. Tanh is not a model of any analog saturator. That said, if we’re trying to stick to functions that roughly emulate physical principles, you can approximate a compressor’s ratio control with a piecewise function that adds saturation without a hard limit.
Clearly you’re doing a lot of plug-in work and have knowledge of the field, but I don’t think I’ve read a single paper or textbook that would agree with your definitions. It really seems like you just have a personal vocabulary and you’re calling people out for not adhering to it.
Im learning a lot. Thank you
Ignore him he's spreading false information.
I am looking for mastering engineer who will clip my carefuly crafted mixes, but no more than 200$ per song.
i'll digitally clip your album for 1000€. i offer the best information loss in town
It seems the "audio engineers" who critized your last video have become very quiet.
indeed lol
I'm sure they'll be back soon, and their comments will be very long.
@@haslo_ The video is very long, so any comment that addresses it has to be reasonably long as well. Here's the short version of mine:
"Good, you're advocating 'soft clipping', kind of like everyone else is, and just using a different term to describe it, which happens to be at odds with the rest of the way the rest of the audio industry has used that term for many decades now" 🙃
my disagreement with the term soft clipping was made very clear in part 1.
Is it me or can any plugin do this by just turning it off? Is this the 'Kings New Clothes' all over again?
yes... saturating and then reversing is completely pointless, it was just a demonstration of how saturation isnt information loss but clipping is.
@@APMasteringThat's because you choose a very soft saturation curve. If you use something with a smaller knee, you'll certainly have information loss in the form of quantization.
@erinburke9711 I totally agree. I did not place enough emphasis on "given sufficient resolution". Given infinite resolution there'd be no loss,, but I agree that's not real world.
@@APMastering Your point is fine since you used tanh. I don't want to give you a hard time.
@@APMastering The fact that you're arguing against hard-clipping is the most important part, and I'm fully onboard.
correction: THIS distortion can be undone
great video as always mate! really informative!
could you explain a bit WHAT clipping is and how clippers work and how it is and it is not just compression? i have trouble finding good info on that
in my last video i show the code for clipping. i don't want to get into too much dsp code on my channel
@@APMastering okay gonna check! thank you!
So how do I just make this in Ableton? lol
there's probably tanh in ableton by i don't know what it's called, i don't use it. but if you have max msp in there you should be able to do it no problems!
is that red hot chili peppers
ha ha, no its just me. I recorded it a few videos ago to demonstrate good vs bad technique
Lady Gaga’s 2nd album vocals
DOPENESS!!!
good vid. except soft clipping is a valid term especially when an adjustable threshold is involved. yes it's just a form of waveshaping with an implicit clip threshold like you demo, however as a tool, an adjustable threshold implies you do want to hard clip above it in a specific place on purpose 🙂. so 'soft clipping' is a good way to think about it.
but tanh can be scaled to have a thresh and is an impassible limit
@@APMastering yes, any waveshaper can. but my point is that if the user is given an adjustable threshold, then clipping is the specific intention of that tool, rather than the byproduct. ie. the name is descriptive of the application, not the dsp.
@@desperateBeauty well i find that saturation is just this. i used to limit stuff by sending it through a tape machine for example. that's saturation by definition. nobody calls that tape soft clipping
wait I'm sorry but when it comes to the speaker issues, aren't they talking about physical clipping? It says "whether the mixer, amplifier, or any other piece of audio equipment clips the signal, Damage can occur without the amp's full output". If they are talking about physical gear causing the clipping i doubt that a loud mix that used smth like classic clipper will cause a pa system to go to shit lol
no they mean ANYTHING that sends a clipped signal to a speaker cone. there's nothing magical about a mixer clipping vs digital clipping
@@APMasteringinteresting i thought it would have something todo with the overloaded signal that is coming in instead of the waveform that it had to reproduce thanks for the info i‘ll look further into this!
@@APMastering wait i‘m still confused if i played a square wave through my speakers for an extended period of time would that break the speakers?
@@7ars471 it could if it is a substantial volume
folding distortion wouldn't work though
sure. and the reverser also required a lot of resolution to accurately remap so it's not a useful process outside of a simple demonstration
Won’t a dc offset remover hi-pass-filter/fft-filter protect the pa from constant power? Just add that before the pa input and no smoke?
no. DC offset filter is just a high pass with a low cut off freq. This will not do much at all. All pass filters are much more effective
@@APMastering isn’t constant voltage == dc? Ac is alternating power == audio signal. A hipass removes oscillation of ~0Hz which is dc offset which is constant power.
@@lazylazy871 yes but the sections of constant power are not
to be more specific, an IIR high pass will make the tops less flat if your Q is gentle because of the phase rotation caused but you achieve the same phase rotation with an all pass. if you use a FIR high pass, so just the filtering without phase impact, there is zero difference to the waveform of a square wave
@@APMastering ah so it’s because the number of samples clipped is low enough and alternating in sign that it is more like a momentary square wave rather than unipolar dc offset. I didn’t think about the basics apparently :D. So now we can conclude that a square wave will also break PAs. Synth bass players better change their waveform to saw then xD
Unfortunately, you have music playing in the background, it's interfering with your speech. I turned off the video.
its -32db... some of the quietest background music on RUclips, but ok.
Haaaaaahhh could you theoretically pull guitar tracks to DI's using this kind of method?
try it out and let me know
This is again fake. Then take off distortion from AC/DC guitars please. You can't fkn redo clipping if its rendered. Unless you are RECORDING in 32bit floating. Also for the clipping drums thing... Clippping does make the drums cut through the song and its vastly different than just adding more volume with a fader. I don't know why you fear losing transients when you are using a daw which have transient designer plugins. With eq and a transient designers i can make even a bass guitar sound like an atomic snare. With limiters/compressors you going to lose transients on drums anyways so its your choice to add them back. Lots of times the drum tracks already have way too much transients and they just poke in too much in the mix.
you didn't watch the video
Vamos por más videos controversiales 🙌👌
isnt this plugin doing the same as a waveshaper, near sqaring it?
wave shaping is just another word for distortion. It is unclear what kind of distortion wave shaping is applying because the term is too vague
sick
Dude, this is primo content. Amazing
Also, for what its worth - an allpass filter (or essentially any filter) will alter peak values when applied after limiting too, not just clipping. Perhaps if anything this is an argument to leave some headroom on the limiter ceiling