FYI as loads of people keep skipping my intros. The mic cut out whilst filming so I had to use the camera audio which isn't great but did the best I could with it 🤓
I pretend to get it, and I want to get it? I think the plugin makers take the Jim Delahunt half the time but hey - we've got to spend our pocket money on something am right? Of course I am! 🤑💰😆
great video. interesting because you showed the value of some functions I don't use. but when it comes to oversampling (or anything else for that matter) I always try to listen carefully rather than analyze visually. certain plugins sound better to me without oversampling despite how it might show on a graph. it's all a give and take.
Yeah totally I find having both approaches really works well. You have a look at what's going on and then see if that correlates with your ears. If it doesn't, it doesn't but i nornally find it helps when listening to how the plugin is actually affecting the source. Helps me focus more. Like the SSL fusion. Soon as I seen the side harmonics in the hardware and none in the plugin I knew where to shift my attention. Gave me perspectives that I was able to transfer to the audio tests and then it's like a door is opened and you can hear what it's actually adding Its a great combo but the ears always have the last word haha
I really like how you are going through all of this..... For me it's not enough just listening to someone say this that and the other, I need to see a visual representation of the explanation. But it has to be absolutey and 100% spelt out to me, but once it has - I'm all over it. .. So I'm realy enjoying all these videos, we're you are working to nail every espect of this subject - cheers Paul 🙂
I think Plugin-Doctor is very handy. It has helped me understand what some of my plugins do in a new way. However it is also important to remember the limitations of what is being tested. Since the sine-wave does not have dynamics, nor affect multiple frequencies you don't get the whole picture of what happens - and the sweep is better, because it affects all frequencies, but not simultanously. Kind of like looking at one leaf getting carried along with the wind is not the same as the dance of 100 leaves, since they also affect each other. And their movement in relationship to each other gets similarly, yet not absolutely, affected by wind and gravity creating some kind of "unison" on a bigger level, while on a smaller level it's still pretty "chaotic". This is a thought of mine that was triggered by a comment from London Acoustics under your Baby Audio video. And I think that's a pretty important observation to keep in mind while looking at these in-depth under-the-hood snapshots of how a plugin behaves under very controlled and limited circumstances that do not at all represent the complexicity of the audiosignals we are using these plugins on.
True. But would you say the sweep still gives a realistic representation? Not absolutely real time bang on but still a good representation? I've asked many developers what they actually use when making the plugins but never had a reply yet 🤣 I know some use plugin doctor but i don't know what else. That seems to be kept very much under wraps from me at this moment in time 😢
@@PaulThird It gives a realistic representation of the sweep. But what does that ACTUALLY represent? It tells you how much the signal gets harmonically affected when a frequency gets hit very shortly in the absence of other frequencies. Wether the frequencies get affected the same way when other frequencies are present at the same time? Wether the frequencies get affected the same way if a frequency is present for a prolongued period of time? What when dynamics are introduced? This will be very plugin-dependent. It all comes down to what the code is. And since I'm not a coder, nor have any idea about what the code of the average plugin actually looks like/how complex they make them - I can't even begin to guess at wether that is a realistic representation of what actually happens on a complex signal. Just keep knocking on doors, one day they'll let you in on all of their secrets! :P
Paul - Very nice job - quick question. Are we supposed to be able to see analysis of currently active plugins inside our channel strips? (running, Logic 11.1.1). PD loads a seperate version of the active strip plugin, outside the channel strip. Was hoping to mix real time with this inside the strip. Don't see anything addressing this on the web. I would appreciate any guidance you may have. Thanks so much.
Chris from Airwindows makes an free ultrasonic filter plugin designed to reduce aliasing. But I don’t really understand how to use it… can’t get my head around the best use! Perhaps after distortion effects, amp sims and certain aliasing plugins to cut it out after it’s generated before it mixes with the rest of the audio.... Maybe thats an cool video idea, kind of fits with the gain staging type series perhaps, getting a more analog sound in the box. Wonder how something like his Channel9 compares to the ones with the fancy interfaces too! :-) Cheers! EDIT: Paul, I was wondering... does having one of these ultrasonic filters after a potentially bad offending plugin kind of solve a lot of the issue, or forego the need for oversampling? Just getting my head around this! Your series is great btw!
Didn't work for me. I got into an argument with a user about it and he was adimant that ultrasonic is not for aliasing and we are misreading Chris's intention for the plugin. After that I just didn't bother. Tried it in different configurations and at different sample rates and had no effect whatsoever 🤷♂️
Paul mate! I’m lovin’ everything you’re doing at the moment - so funny that now all these big shot UA-camrs are turning their heads your way, you’re quickly earning the respect of being the only guy who actually dares to explain all this stuff to a growing audience of plug-in enthusiasts who would be otherwise totally baffled by this stuff!! Don’t hate me mate, but I do have a question for you - something that was sparked in my mind when a White Sea Studio video mentioned it in-passing whilst talking about how the printed audio master is affected by the working sample rate of the actual DAW project, which can have unexpected results if there’s a mismatch in the sample rate between plugin real-time playback and rendering… So it’s this: It seems that certain plug-ins - Acustica’s Aquas being amongst them - apply a very steep low-pass filter to the output - kinda the opposite end of what I believe a “DC blocking” filter is - so this is certainly to help mitigate the unwanted aliasing of insufficiently over-sampled audio - but my question is simply “how and why??” I.e. Should I applying a super-steep low-pass filter to the input of a clipper plugin for example? Should I be manually putting a filter like this on anything for that matter? Furthermore, I’m confused about Acustica’s method of dealing with aliasing on their Aqua plug-ins in general? Are they adding a low-pass filter *instead* of using over-sampling?? And are they adding it *above* the nyquist ceiling of the max-frequency (like above 22kHz for a 44.1kHz sample rate), or is it really just a down-and-dirty way to stop aliasing frequencies mirroring down on themselves again?? I can’t be the only person who wonders this… I try and take this stuff with as big a pinch of salt as possible, but once I found out that an alarming amount of my EQ plug-ins have cramping in the top end, I’ve spotted this secret low-pass filter being used more and more!! Bigs up champ
The thing you have to remember about aliasing is that a lpf at 20k will only cut off harmonics made from frequencies at 20k & above. You'll still have orders of harmonics made at frequencies below 20k. That's why I do aliasing tests at 10k as it can give a realistic result as most audio sources will have information around 10k. The harmonics made depend on what frequencies are actually in the source. A lpf won't cut off harmonics made by frequencies below the lpf. It'll just cut off the frequencies which means they also won't create harmonics. Increasing niquist is the best way to reduce aliasing as it means the computer can generate those frequencies and harmonics and thus reduce the amount of harmonics bouncing back. The lpf in conjuction with increasing niquist makes sense to me as in my head essentially what its doing is increasing niquist to the point where the orders of harmonics made from frequencies below 20 or 22k end or become inaudible. The truth is, a lot of hardware actually is quite low in THD, as is many acustica plugins. If you don't make a lot of audible high order harmonics then you won't get any audible aliasing. I think there is some anti aliasing filters going on in conjuction with low THD in the Acustica plugs why by default will halt the creation of most aliasing. Oversampling is increasing niquist which is the end of your frequency range. Lpf just cuts off frequencies. A combination of both is what we normally see in the plugins that create the full THD where tone empire took the approach of only adding in enough saturation per oversampling option so it made just enough high order harmonics surpassing nuquist without coming back as audible aliasing. You can lpf out all the frequencies until they stop making harmonics that surpass niquist but you may be looking at like a 4-5k lpf in some cases which isn't realistic in most cases. Increasing your sample rate with a rich harmonic plugin is the best way to get the full audible frequency range without adding in lots of audible aliasing.
Even worth checking harmonics at 10hz as it's the orders of harmonics you need to keep in mind. Transformer harmonics are normally very low down the frequency spectrum but the order of harmonics stretches into our audible range dependent on the strength of those low order harmonics. Do you need to worry about aliasing that low.. No, but it tells you why you are percieving a bit more overall weight independent from frequency response
such a gem! i’m liking my processing but using headphones for now i “clip” my print track but my peaks aren’t over 😭 my ITB is 82DBVU when i open it up using the Gsat in the sides for harmonics on verbs and delay !!!! 🔥😳
I dunno why evertime I try to change the oversampling anything past 4x and it crashes my DAW.. but I can max out other plugins just fine doing x8 or even higher
Another good whan Paul, I'm trying to understand all of what your covering as a novice in the house, a meer hoose moose recorder I am. I'm proud & glad a Scottish lad had came on the scene of audio, I think theres another dude but your funnier & ye swear like normal people wich takes balls, glad your one of the troops in that respect Paul. Whether your right or wrong I have no idea but I do learn things to jump about my head and may make sense (Eurica) as I progress. I know as I was a camera /editing dude for years it can take a whole day /s to make the vids so I appreciate it matey cheers - fuck knows who gave you a thumbs Doon? Must be that EX that canny handle ye doing well in yer wee life 😜 Joost kidding its a fanny who disagrees harmonically with a niquist disgruntlement (ye get them) I'm just happy your oot there dane it.. (I got very Scottish there) Cheers Paul, I hope its okay to ask questions about Recording in my wee flat, I've got good gear & good plugins, it's the 'how do I make my songs sound like a record / panning tracks / mono or stereo order etc? Hae a guid wknd Paul.. ✊🏾✊🏻😎
@@PaulThird I've just tapped a lenny me birds mic as she's away the noo.. Brand new Neuman.. If I break it she'll kill me.. I won't, I'm extra para wae other peoples gear.. Here, what do you think I should put on that wee song I posted for ye on Twatter? I wrote it in an hour last week, I've got loads.. I can write about 3 songs a day when I'm on a roll, I dnt hang about with any musicians so I'm doing it all myself & I'm fkd with all the effort, if ye have any ideas as to what you think it should be like please let me know man? I just want to make my songs sound good so my daughter will stop saying 'Da, your no gonny be a rock star' when I tell her 'I fkn am & when I'm playing Glasto your no getting a back stage pass ya wee shite'... Ye never give up on yer dreams Paul.. Never! ✊🏻🔉🌎
I just realised I dinna get twitter notifications haha had like 45 mentions and had no idea 😅 My advice is just to make the music, do the best you can. Reference it against other tracks. Mix it the best you can. Release it. Learn from the experience and make good of the mistakes you learned from the last record. Try and improve with every record. That's what I'm doing anyway
and if 80 db is the threshold where harmonics can be heard what is the point with aa plugs.... .. their orders barely touch -100 and if you raise gain to raise up the harmonics to be above -80 aliasing is introduced... i just got done testing gain-station and as soon as you take the clip to +12 a ridiculous amount of aliasing comes in.... but just a smidgen under -12 and it goes away..
Don't own any AA plugs. barely touching -100 seems odd to me though. Are you talking about the noise-floor or the actual harmonic peaks? I don't mean to be disrespectful - I'm just trying to make sure we're talking about the same thing.
Have you tested at different frequencies instead of 1k and tried using the compressors and stuff? When I printed sweeps of a lot of hardware they were actually pretty clean tbh depends on the gear
@@IamtheI well "color" me interested. Do you mind to share what AA plugs you are specifically talking about? PS: I wouldn't say -80 Is inaudible.. but it will be very, very, very subtle. Also stuff adds up. Add -80 dB of reverb to one scource and you are unlikely to hear it. Add it to all and you just might notice. But saying something is subtle, when it's next to inaudible is close to crowning bullshit.
does it matter what device is selected in setting for sounds?? for instance should i have it set to my windows audio which is feeding my bose companinon or should i have it feed into the X8p's asio... i have been noticing no difference to worst results swithching between 44 and 96..... tested a bunch of plugs and the aiiasing is the same in both sample rates... do i need to changet the sample rate in my interface or windows audio as well// :)
@@PaulThird not sure how that would work... i mean i have my uad interfaces connected to my computer..... in PD's there is section called show audio hardware settngs.. not sure if those settings have anything to do with just testing a plugin..... im not trying to connect actual hardware though (ie analog eq, comp, pre etc..... :-)
Nah hardware settings is for connecting hardware I'm pretty sure. Well that's all I've ever used it for. I've always just opened plugin doctor, loaded a plugin and that's it Might be an issue knowing what interface or asio to use though so maybe best to try a few things out Selecting your output or input in hardware settings Also what is tone frequency to check aliasing?
Three observations... 1. DDMF needs to create comprehensive tutorials - I can't think of anybody who has released a measurement tool without detailed instructions of how to interpret their measurements. 2. I think the "fundamental sweep" is in the name... the trace line is showing us the fundamental only and how it's level is affected by non-linearities (educated guess as I'm not at my music computer!). 3. The oversampling observation on Tone Empire "up for debate"... all we should be seeing is the reduction of aliasing - this change of saturation at higher oversampling may be unique to Tone Empire.
@@PaulThird If I didn't know better... I'd say that DDMF doesn't want us to fully understand their measurements as *somebody* is likely to uncover anomolies within DDMF's own audio plugins. 😉
I’m not an audio software programmer, so there’s a flaw in this logic I don’t understand: why can’t aliasing be stopped by some sort of check on where the generated harmonics will end up (before Nyquist or after) So the plug-in does the math to generate harmonics, then there’s a second step that checks where all of them are in the frequency spectrum and omits everything passed say 18K at 44.1. But just the harmonics are omitted so they don’t fold back. Is the problem simply that it’d be too CPU and latency intensive to make sense? Even if possible, I see no way it wouldn’t add latency. Anyway, I geek out on this stuff too. But I’m supposed to be working right now so back at it!! 😂
You can't filter out frequencies that Nyquist can't cover: The fold-back distortion already happened before you can apply any filtering. I could imagine it could be possible (in theory at least) to do a frequency dependent saturation: Say, for example you have a simple hard-clipper that adds odd harmonics, and you want to prevent those harmonics to fold back. Then only clip those frequencies where the resulting harmonics wouldn't get outside of Nyquist (easy to predict with a simple hard-clipper) and leave the rest alone. Maybe this is possible by dividing the input signal into many narrow frequency bands, or maybe something could be done using a Fourier transform. However, it would be waaaaaay more CPU consuming than simply applying oversampling, and I suspect that in practice it's pretty much impossible to make it work perfectly anyway so the result wouldn't be any better than oversampling. But anyway, I haven't tried something like this myself, so maybe I'm forgetting something. But in the end, oversampling is a relatively cheap way to properly prevent/reduce aliasing. And in my own experience aliasing is a relatively small issue that's usually hardly audible in a full mix (if at all), especially when the channels with more heavy saturation are oversampled. Perhaps a more practical solution would be if the DAW would apply a very brute-force amount of oversampling for every plugin when rendering the mix, to get rid of any remaining aliasing.
@@jaydy71 this is kind of what I figured, that even if there is a way it’d be way too CPU expensive, particularly when oversampling is effective, not terribly expensive, and I imagine easy to implement most of the time. DAW’s including an “upsample on render” option would be fabulous. Maybe it could cause problems with some VSTs?
@@mrnelsonius5631 Regarding the 'upsample on render' option, it's something I'd really like. Actually it's a feature I proposed to Steinberg for Cubase (next to a feature to oversample any plugin). I think it would actually be rather easy to implement: Just oversample before the signal hits the plugin, tell the plugin the oversampled sample-rate, apply the anti-aliasing filter and then downsample the output. I don't think it would be a problem for any VST because from the plugin's perspective it wouldn't be any different than having the whole project running at a higher sample-rate. I believe there are actually some 'meta-plugins' that can do that, but I haven't looked into those yet. I suppose a potential issue with the 'upsample on render' option is that what's being rendered will sound slightly different than what you've actually mixed (perhaps some of the aliasing was part of the sound you were going for, or something like that), but I figure in most cases it will just sound better.
@@jaydy71 yeah, and there is also a quirk in how a lot of plugins handle the project sample rate. Many essentially scan for the sample rate at the moment of being instantiated in the DAW. Such as many convolution plugins: they operate differently depending on the project sample rate, and a mismatch can cause audio issues. Acustica had a problem for a long time where, if you started the project at one sample rate then changed to a higher one before export, all your EQ curves would be totally offset. It would destroy your mix. I *think* they eventually fixed it but I could see other plugins struggling with an external sample rate change too. The thing about plugins is that they all work in very different ways. Just like aliasing, developers can handle it quite differently depending on the quirks of their own processing methods. It’s still a good idea, and if something needs to change in the VST format to accommodate sample rate changes on the fly, it should
there isn't any adc that doesn't add a high pass filter at 22.5 kHz (nyquist) that's wht they're trying to replicate using anti-aliasing filter but for aliaisng. at supersonic frequencies above 22 Khz.
@@PaulThird yes they do but having it lower than 48 kHz can screw up with the high end on an adc for example on 44 kHz. 96 kHz yields the best result in retaining top end on a ADC. Most plugin that adds an anti-aliasing filter removes it when the project is run at 96 kHz or higher so at this time the daw does the job but most daw have very bad sample rate conversion which leaves some of the aliasing in the top end there's three ways you could fix that 1) it's to use a daw with a good anti-aliasing filter when converting to a lower sample rate from 96. 2) export the file at 96 and convert using dbpoweramp or izotope rx and some others that yields very good sample conversion and antia-aliasing filter. 3) use tools like ddmf metaplugin on 2x oversample for offline boucing or airwindows ultrasonic at the end of your analog modelled plugin chain when your running your project at 96k.
I guess I don't rightly care what a plugin is doing inside. If my EQ says one thing but delivers something slightly "not exactly that", this is the nature of a real Neve or SSL. What I do care about is if it delivers the results the Story of the Song I am mixing needs. That said there are times I do want to know what a process is doing so I know how to do those things myself, my way. I have seen a vid showing that some plugins making grand claims are way off on what they say. That is plain uncool - Emperor's New Clothes. :-)
That's how I use it. Once I know what it's doing I see if I can recreate it myself. Saves a lot of money that way for me and gives me more insight to what it's actually doing so i know if I'm possibly buying something that I've essentially already got. Or on the other hand gives me better context to listen to what it's adding that I might not be aware of
Not a fan of what Tone-Empire does there.. oversampling should not increase the harmonics imo. To me oversampling should be simply keep the plugin from aliasing. Very interesting that this happens though. Guess one should be careful with using plugin-inherent-oversampling-options.
And that's why i use plugin doctor haha I feel understanding what's going on under the hood gives you way better understanding of how to use it the way you need or how not to use it I don't believe in magic any more 😢 🤣
@@PaulThird "witchcraft and technology - get mistaken commonly". I promise you this: There's still technology out there, that will have you believe in witchcraft. Or maybe it's actually witchcraft. I don't know. But I agree that Plugin Doctor helps using the plugins more conciously/better informed/less intuitively. I think that's a good thing for an engineer.
i think its a fine compromise for the betterment of workflow, and i think its actually quite clever as it is much easier to dial in saturation "til it sounds good," and feeling somewhat comfortable knowing that no aliasing was introduced in the process, than trying to listen to the aliasing yourself and deciding whether oversampling is needed or not, it is definitely still a compromise but thats just my view!
I agree. I think it's quite clever especially as they have stated that this is a unique hybrid plugin of its own so it not forced like other plugins to give the modeled sound of the hardware at all settings. They can do what they like and they've chosen to give you more saturation options which I think is pretty cool. Is it a bummer that you have to use X8 for full saturation.. Little bit but it's still giving you options without worrying too much about aliasing. If it was a straight up analog emulation though I'd be saying different but I like the idea of these hybrid plugins. not limiting ourselves and giving us different options to play about with
FYI as loads of people keep skipping my intros. The mic cut out whilst filming so I had to use the camera audio which isn't great but did the best I could with it 🤓
People have no clue of anything. hahaha!!!
Hahaha well we can still try 😂
I pretend to get it, and I want to get it? I think the plugin makers take the Jim Delahunt half the time but hey - we've got to spend our pocket money on something am right? Of course I am! 🤑💰😆
I just presumed people who are watching get it? Back tae the drawing board lads 🙄
Its been a quiet video haha actually one of the smallest comment sections I've had. Maybe it's just a bit much haha
Man u r the coolest. Dude really! You are the goat when it comes to revealing plugin bull bull.
🤜🤛
great video. interesting because you showed the value of some functions I don't use. but when it comes to oversampling (or anything else for that matter) I always try to listen carefully rather than analyze visually. certain plugins sound better to me without oversampling despite how it might show on a graph. it's all a give and take.
Yeah totally I find having both approaches really works well. You have a look at what's going on and then see if that correlates with your ears. If it doesn't, it doesn't but i nornally find it helps when listening to how the plugin is actually affecting the source. Helps me focus more. Like the SSL fusion. Soon as I seen the side harmonics in the hardware and none in the plugin I knew where to shift my attention. Gave me perspectives that I was able to transfer to the audio tests and then it's like a door is opened and you can hear what it's actually adding
Its a great combo but the ears always have the last word haha
Good video Paul, appreciate the effort that you put in. Keep 'em coming
🤓🤓
I really like how you are going through all of this..... For me it's not enough just listening to someone say this that and the other, I need to see a visual representation of the explanation. But it has to be absolutey and 100% spelt out to me, but once it has - I'm all over it.
.. So I'm realy enjoying all these videos, we're you are working to nail every espect of this subject - cheers Paul 🙂
🤓🤓
Thanks for this incredible tutorial! Love how you broke it down!!
🤓🤓
AWESOME work dude..!!!
At last you did it and looks great.
Thanks for your GREAT work.
Bless..!!
🤓🤓
finally .. starting to make sum sense my dude!!
🙌🙌
I think Plugin-Doctor is very handy. It has helped me understand what some of my plugins do in a new way. However it is also important to remember the limitations of what is being tested. Since the sine-wave does not have dynamics, nor affect multiple frequencies you don't get the whole picture of what happens - and the sweep is better, because it affects all frequencies, but not simultanously. Kind of like looking at one leaf getting carried along with the wind is not the same as the dance of 100 leaves, since they also affect each other. And their movement in relationship to each other gets similarly, yet not absolutely, affected by wind and gravity creating some kind of "unison" on a bigger level, while on a smaller level it's still pretty "chaotic".
This is a thought of mine that was triggered by a comment from London Acoustics under your Baby Audio video. And I think that's a pretty important observation to keep in mind while looking at these in-depth under-the-hood snapshots of how a plugin behaves under very controlled and limited circumstances that do not at all represent the complexicity of the audiosignals we are using these plugins on.
True. But would you say the sweep still gives a realistic representation? Not absolutely real time bang on but still a good representation?
I've asked many developers what they actually use when making the plugins but never had a reply yet 🤣
I know some use plugin doctor but i don't know what else. That seems to be kept very much under wraps from me at this moment in time 😢
@@PaulThird It gives a realistic representation of the sweep. But what does that ACTUALLY represent? It tells you how much the signal gets harmonically affected when a frequency gets hit very shortly in the absence of other frequencies. Wether the frequencies get affected the same way when other frequencies are present at the same time? Wether the frequencies get affected the same way if a frequency is present for a prolongued period of time? What when dynamics are introduced?
This will be very plugin-dependent. It all comes down to what the code is. And since I'm not a coder, nor have any idea about what the code of the average plugin actually looks like/how complex they make them - I can't even begin to guess at wether that is a realistic representation of what actually happens on a complex signal.
Just keep knocking on doors, one day they'll let you in on all of their secrets! :P
OMG! This is pure gold
🤓🤓
Great! Just what I needed!!
🤓🤓
Paul - Very nice job - quick question. Are we supposed to be able to see analysis of currently active plugins inside our channel strips? (running, Logic 11.1.1). PD loads a seperate version of the active strip plugin, outside the channel strip. Was hoping to mix real time with this inside the strip. Don't see anything addressing this on the web. I would appreciate any guidance you may have. Thanks so much.
You would need to use ddmf metaplugin to do that in doctor
Is there an alternative to Plug-in Doctor for Pro Tools?
Not that I know of
Chris from Airwindows makes an free ultrasonic filter plugin designed to reduce aliasing. But I don’t really understand how to use it… can’t get my head around the best use! Perhaps after distortion effects, amp sims and certain aliasing plugins to cut it out after it’s generated before it mixes with the rest of the audio.... Maybe thats an cool video idea, kind of fits with the gain staging type series perhaps, getting a more analog sound in the box. Wonder how something like his Channel9 compares to the ones with the fancy interfaces too! :-) Cheers!
EDIT: Paul, I was wondering... does having one of these ultrasonic filters after a potentially bad offending plugin kind of solve a lot of the issue, or forego the need for oversampling? Just getting my head around this! Your series is great btw!
Didn't work for me. I got into an argument with a user about it and he was adimant that ultrasonic is not for aliasing and we are misreading Chris's intention for the plugin.
After that I just didn't bother. Tried it in different configurations and at different sample rates and had no effect whatsoever 🤷♂️
Great video! Would you be able to do a video on how to gain stage into these plugins correctly with tracks in a mix to maximize the THD and curves?
ua-cam.com/video/0-2zTMjPAqk/v-deo.html
ua-cam.com/video/w0DTz1XGDJo/v-deo.html
Paul mate! I’m lovin’ everything you’re doing at the moment - so funny that now all these big shot UA-camrs are turning their heads your way, you’re quickly earning the respect of being the only guy who actually dares to explain all this stuff to a growing audience of plug-in enthusiasts who would be otherwise totally baffled by this stuff!!
Don’t hate me mate, but I do have a question for you - something that was sparked in my mind when a White Sea Studio video mentioned it in-passing whilst talking about how the printed audio master is affected by the working sample rate of the actual DAW project, which can have unexpected results if there’s a mismatch in the sample rate between plugin real-time playback and rendering…
So it’s this: It seems that certain plug-ins - Acustica’s Aquas being amongst them - apply a very steep low-pass filter to the output - kinda the opposite end of what I believe a “DC blocking” filter is - so this is certainly to help mitigate the unwanted aliasing of insufficiently over-sampled audio - but my question is simply “how and why??”
I.e. Should I applying a super-steep low-pass filter to the input of a clipper plugin for example? Should I be manually putting a filter like this on anything for that matter?
Furthermore, I’m confused about Acustica’s method of dealing with aliasing on their Aqua plug-ins in general? Are they adding a low-pass filter *instead* of using over-sampling?? And are they adding it *above* the nyquist ceiling of the max-frequency (like above 22kHz for a 44.1kHz sample rate), or is it really just a down-and-dirty way to stop aliasing frequencies mirroring down on themselves again?? I can’t be the only person who wonders this…
I try and take this stuff with as big a pinch of salt as possible, but once I found out that an alarming amount of my EQ plug-ins have cramping in the top end, I’ve spotted this secret low-pass filter being used more and more!!
Bigs up champ
The thing you have to remember about aliasing is that a lpf at 20k will only cut off harmonics made from frequencies at 20k & above. You'll still have orders of harmonics made at frequencies below 20k. That's why I do aliasing tests at 10k as it can give a realistic result as most audio sources will have information around 10k. The harmonics made depend on what frequencies are actually in the source.
A lpf won't cut off harmonics made by frequencies below the lpf. It'll just cut off the frequencies which means they also won't create harmonics.
Increasing niquist is the best way to reduce aliasing as it means the computer can generate those frequencies and harmonics and thus reduce the amount of harmonics bouncing back.
The lpf in conjuction with increasing niquist makes sense to me as in my head essentially what its doing is increasing niquist to the point where the orders of harmonics made from frequencies below 20 or 22k end or become inaudible.
The truth is, a lot of hardware actually is quite low in THD, as is many acustica plugins. If you don't make a lot of audible high order harmonics then you won't get any audible aliasing.
I think there is some anti aliasing filters going on in conjuction with low THD in the Acustica plugs why by default will halt the creation of most aliasing.
Oversampling is increasing niquist which is the end of your frequency range. Lpf just cuts off frequencies.
A combination of both is what we normally see in the plugins that create the full THD where tone empire took the approach of only adding in enough saturation per oversampling option so it made just enough high order harmonics surpassing nuquist without coming back as audible aliasing.
You can lpf out all the frequencies until they stop making harmonics that surpass niquist but you may be looking at like a 4-5k lpf in some cases which isn't realistic in most cases.
Increasing your sample rate with a rich harmonic plugin is the best way to get the full audible frequency range without adding in lots of audible aliasing.
@@PaulThird that whole Q&A was just wonderful... I am officially a geek too..
The one I made or the twitch stream?
@@PaulThird explanation greatly appreciated!
When will you do an episode on inter modulation distortion?
When I have a better understanding of it 😅 my understanding of it is pretty basic but I do plan on looking more into it
I learned a lot. Thanks!
🤓🤓
Really helpful tutorial, Paul.
I'm wondering if 50Hz for checking harmonics content is just a random choice or there is some logics behind it?
Even worth checking harmonics at 10hz as it's the orders of harmonics you need to keep in mind. Transformer harmonics are normally very low down the frequency spectrum but the order of harmonics stretches into our audible range dependent on the strength of those low order harmonics. Do you need to worry about aliasing that low.. No, but it tells you why you are percieving a bit more overall weight independent from frequency response
@@PaulThird Got it, thank you!
Paul, do you still use Acustica stuff? Or are you looking for alternatives?
Yeah I still use a lot of AA but I'm toning it down and using more pre's than channel strips and stuff. Just trying to find a balance
Great video again mate can you do a test of Total Eq by Hornet?
I've been told hornets very good quality 🤓
@@PaulThirdTotal eq has Analog and Saturation modes I done some tests in plugin doctor and there is no alliasing I think
such a gem! i’m liking my processing but using headphones for now i “clip” my print track but my peaks aren’t over 😭 my ITB is 82DBVU when i open it up using the Gsat in the sides for harmonics on verbs and delay !!!! 🔥😳
Hey Paul,
How do you load Waves plugins in plugin doctor? Cause they don't have .component files
You load vst & vst 3
I dunno why evertime I try to change the oversampling anything past 4x and it crashes my DAW.. but I can max out other plugins just fine doing x8 or even higher
Metaplugin?
@@PaulThird tone empire plugins
That's odd, doesn't happen with me 🤔
Hello, What version of the Audient is that?
Id44
Another good whan Paul, I'm trying to understand all of what your covering as a novice in the house, a meer hoose moose recorder I am. I'm proud & glad a Scottish lad had came on the scene of audio, I think theres another dude but your funnier & ye swear like normal people wich takes balls, glad your one of the troops in that respect Paul. Whether your right or wrong I have no idea but I do learn things to jump about my head and may make sense (Eurica) as I progress. I know as I was a camera /editing dude for years it can take a whole day /s to make the vids so I appreciate it matey cheers - fuck knows who gave you a thumbs Doon? Must be that EX that canny handle ye doing well in yer wee life 😜 Joost kidding its a fanny who disagrees harmonically with a niquist disgruntlement (ye get them) I'm just happy your oot there dane it.. (I got very Scottish there) Cheers Paul, I hope its okay to ask questions about Recording in my wee flat, I've got good gear & good plugins, it's the 'how do I make my songs sound like a record / panning tracks / mono or stereo order etc?
Hae a guid wknd Paul.. ✊🏾✊🏻😎
Good gear and good plugins.. Sounds like a good weekend haha 😂
@@PaulThird I've just tapped a lenny me birds mic as she's away the noo.. Brand new Neuman.. If I break it she'll kill me.. I won't, I'm extra para wae other peoples gear..
Here, what do you think I should put on that wee song I posted for ye on Twatter? I wrote it in an hour last week, I've got loads.. I can write about 3 songs a day when I'm on a roll, I dnt hang about with any musicians so I'm doing it all myself & I'm fkd with all the effort, if ye have any ideas as to what you think it should be like please let me know man? I just want to make my songs sound good so my daughter will stop saying 'Da, your no gonny be a rock star' when I tell her 'I fkn am & when I'm playing Glasto your no getting a back stage pass ya wee shite'... Ye never give up on yer dreams Paul.. Never! ✊🏻🔉🌎
I just realised I dinna get twitter notifications haha had like 45 mentions and had no idea 😅
My advice is just to make the music, do the best you can. Reference it against other tracks. Mix it the best you can. Release it.
Learn from the experience and make good of the mistakes you learned from the last record. Try and improve with every record.
That's what I'm doing anyway
@@PaulThird just merr pish tae go thru mate... 🤘🏽🙄
Is the DDMF website fixed yet? I tried a week or so ago to possibly get Plugin Doctor but the site had the "couldn't reach, try again later" message.
Worked for me the other day 🤓
and if 80 db is the threshold where harmonics can be heard what is the point with aa plugs.... .. their orders barely touch -100 and if you raise gain to raise up the harmonics to be above -80 aliasing is introduced... i just got done testing gain-station and as soon as you take the clip to +12 a ridiculous amount of aliasing comes in.... but just a smidgen under -12 and it goes away..
Don't own any AA plugs. barely touching -100 seems odd to me though. Are you talking about the noise-floor or the actual harmonic peaks? I don't mean to be disrespectful - I'm just trying to make sure we're talking about the same thing.
Have you tested at different frequencies instead of 1k and tried using the compressors and stuff? When I printed sweeps of a lot of hardware they were actually pretty clean tbh depends on the gear
@@akagerhard also the same thing happens if i leave the clipper alone and do the same thing at +12 on the pre dial
@@akagerhard hey partner :-) ... im tallking about the harmonic peaks...
@@IamtheI well "color" me interested. Do you mind to share what AA plugs you are specifically talking about?
PS: I wouldn't say -80 Is inaudible.. but it will be very, very, very subtle. Also stuff adds up. Add -80 dB of reverb to one scource and you are unlikely to hear it. Add it to all and you just might notice. But saying something is subtle, when it's next to inaudible is close to crowning bullshit.
does it matter what device is selected in setting for sounds?? for instance should i have it set to my windows audio which is feeding my bose companinon or should i have it feed into the X8p's asio... i have been noticing no difference to worst results swithching between 44 and 96..... tested a bunch of plugs and the aiiasing is the same in both sample rates... do i need to changet the sample rate in my interface or windows audio as well// :)
Are you talking about connecting hardware to plugin doctor?
@@PaulThird not sure how that would work... i mean i have my uad interfaces connected to my computer..... in PD's there is section called show audio hardware settngs.. not sure if those settings have anything to do with just testing a plugin..... im not trying to connect actual hardware though (ie analog eq, comp, pre etc..... :-)
Nah hardware settings is for connecting hardware I'm pretty sure. Well that's all I've ever used it for.
I've always just opened plugin doctor, loaded a plugin and that's it
Might be an issue knowing what interface or asio to use though so maybe best to try a few things out Selecting your output or input in hardware settings
Also what is tone frequency to check aliasing?
@@PaulThird 9999k... i posted a link with a pic showing the 3 test..... its not appearing here anymore though /.... :-)
🤓🤓🤓
Remember to check out my autism channel if you want to learn more about my life 🤓🤓
Three observations... 1. DDMF needs to create comprehensive tutorials - I can't think of anybody who has released a measurement tool without detailed instructions of how to interpret their measurements. 2. I think the "fundamental sweep" is in the name... the trace line is showing us the fundamental only and how it's level is affected by non-linearities (educated guess as I'm not at my music computer!). 3. The oversampling observation on Tone Empire "up for debate"... all we should be seeing is the reduction of aliasing - this change of saturation at higher oversampling may be unique to Tone Empire.
It's not easy is it haha lots of guess work 😅
@@PaulThird If I didn't know better... I'd say that DDMF doesn't want us to fully understand their measurements as *somebody* is likely to uncover anomolies within DDMF's own audio plugins. 😉
Would actually make sense haha
I’m not an audio software programmer, so there’s a flaw in this logic I don’t understand: why can’t aliasing be stopped by some sort of check on where the generated harmonics will end up (before Nyquist or after) So the plug-in does the math to generate harmonics, then there’s a second step that checks where all of them are in the frequency spectrum and omits everything passed say 18K at 44.1. But just the harmonics are omitted so they don’t fold back. Is the problem simply that it’d be too CPU and latency intensive to make sense? Even if possible, I see no way it wouldn’t add latency. Anyway, I geek out on this stuff too. But I’m supposed to be working right now so back at it!! 😂
There will be a reason. I always think 'if it sounds easy to do and it's not been done'.. Its not easy to do haha
You can't filter out frequencies that Nyquist can't cover: The fold-back distortion already happened before you can apply any filtering.
I could imagine it could be possible (in theory at least) to do a frequency dependent saturation: Say, for example you have a simple hard-clipper that adds odd harmonics, and you want to prevent those harmonics to fold back. Then only clip those frequencies where the resulting harmonics wouldn't get outside of Nyquist (easy to predict with a simple hard-clipper) and leave the rest alone. Maybe this is possible by dividing the input signal into many narrow frequency bands, or maybe something could be done using a Fourier transform.
However, it would be waaaaaay more CPU consuming than simply applying oversampling, and I suspect that in practice it's pretty much impossible to make it work perfectly anyway so the result wouldn't be any better than oversampling.
But anyway, I haven't tried something like this myself, so maybe I'm forgetting something.
But in the end, oversampling is a relatively cheap way to properly prevent/reduce aliasing. And in my own experience aliasing is a relatively small issue that's usually hardly audible in a full mix (if at all), especially when the channels with more heavy saturation are oversampled.
Perhaps a more practical solution would be if the DAW would apply a very brute-force amount of oversampling for every plugin when rendering the mix, to get rid of any remaining aliasing.
@@jaydy71 this is kind of what I figured, that even if there is a way it’d be way too CPU expensive, particularly when oversampling is effective, not terribly expensive, and I imagine easy to implement most of the time. DAW’s including an “upsample on render” option would be fabulous. Maybe it could cause problems with some VSTs?
@@mrnelsonius5631 Regarding the 'upsample on render' option, it's something I'd really like. Actually it's a feature I proposed to Steinberg for Cubase (next to a feature to oversample any plugin). I think it would actually be rather easy to implement: Just oversample before the signal hits the plugin, tell the plugin the oversampled sample-rate, apply the anti-aliasing filter and then downsample the output.
I don't think it would be a problem for any VST because from the plugin's perspective it wouldn't be any different than having the whole project running at a higher sample-rate. I believe there are actually some 'meta-plugins' that can do that, but I haven't looked into those yet.
I suppose a potential issue with the 'upsample on render' option is that what's being rendered will sound slightly different than what you've actually mixed (perhaps some of the aliasing was part of the sound you were going for, or something like that), but I figure in most cases it will just sound better.
@@jaydy71 yeah, and there is also a quirk in how a lot of plugins handle the project sample rate. Many essentially scan for the sample rate at the moment of being instantiated in the DAW. Such as many convolution plugins: they operate differently depending on the project sample rate, and a mismatch can cause audio issues. Acustica had a problem for a long time where, if you started the project at one sample rate then changed to a higher one before export, all your EQ curves would be totally offset. It would destroy your mix. I *think* they eventually fixed it but I could see other plugins struggling with an external sample rate change too. The thing about plugins is that they all work in very different ways. Just like aliasing, developers can handle it quite differently depending on the quirks of their own processing methods. It’s still a good idea, and if something needs to change in the VST format to accommodate sample rate changes on the fly, it should
Distressor vs Arousor vs Mike-e
I've done distressor vs mike-e. That was enough for me haha
get the mike-e and you will be very happy
Easy for some to say 😜
@@PaulThird but the opto mode?? The distressor have a opto mode, the arousor too
there isn't any adc that doesn't add a high pass filter at 22.5 kHz (nyquist) that's wht they're trying to replicate using anti-aliasing filter but for aliaisng. at supersonic frequencies above 22 Khz.
So does every ADC cut at 22.5k regardless of sample rate?
@@PaulThird yes they do but having it lower than 48 kHz can screw up with the high end on an adc for example on 44 kHz. 96 kHz yields the best result in retaining top end on a ADC. Most plugin that adds an anti-aliasing filter removes it when the project is run at 96 kHz or higher so at this time the daw does the job but most daw have very bad sample rate conversion which leaves some of the aliasing in the top end there's three ways you could fix that 1) it's to use a daw with a good anti-aliasing filter when converting to a lower sample rate from 96. 2) export the file at 96 and convert using dbpoweramp or izotope rx and some others that yields very good sample conversion and antia-aliasing filter. 3) use tools like ddmf metaplugin on 2x oversample for offline boucing or airwindows ultrasonic at the end of your analog modelled plugin chain when your running your project at 96k.
I guess I don't rightly care what a plugin is doing inside. If my EQ says one thing but delivers something slightly "not exactly that", this is the nature of a real Neve or SSL. What I do care about is if it delivers the results the Story of the Song I am mixing needs. That said there are times I do want to know what a process is doing so I know how to do those things myself, my way. I have seen a vid showing that some plugins making grand claims are way off on what they say. That is plain uncool - Emperor's New Clothes. :-)
That's how I use it. Once I know what it's doing I see if I can recreate it myself. Saves a lot of money that way for me and gives me more insight to what it's actually doing so i know if I'm possibly buying something that I've essentially already got. Or on the other hand gives me better context to listen to what it's adding that I might not be aware of
I prefer HELLLLL LOOOOO EVERRRY ONE! this one was weak! :P
It depends on my mood that week haha
what’s your IG Paul?!
@paulthird_uk
Not a fan of what Tone-Empire does there.. oversampling should not increase the harmonics imo. To me oversampling should be simply keep the plugin from aliasing. Very interesting that this happens though. Guess one should be careful with using plugin-inherent-oversampling-options.
And that's why i use plugin doctor haha I feel understanding what's going on under the hood gives you way better understanding of how to use it the way you need or how not to use it
I don't believe in magic any more 😢 🤣
@@PaulThird "witchcraft and technology - get mistaken commonly". I promise you this: There's still technology out there, that will have you believe in witchcraft. Or maybe it's actually witchcraft. I don't know.
But I agree that Plugin Doctor helps using the plugins more conciously/better informed/less intuitively. I think that's a good thing for an engineer.
100%
i think its a fine compromise for the betterment of workflow, and i think its actually quite clever
as it is much easier to dial in saturation "til it sounds good," and feeling somewhat comfortable knowing that no aliasing was introduced in the process, than trying to listen to the aliasing yourself and deciding whether oversampling is needed or not, it is definitely still a compromise
but thats just my view!
I agree. I think it's quite clever especially as they have stated that this is a unique hybrid plugin of its own so it not forced like other plugins to give the modeled sound of the hardware at all settings. They can do what they like and they've chosen to give you more saturation options which I think is pretty cool. Is it a bummer that you have to use X8 for full saturation.. Little bit but it's still giving you options without worrying too much about aliasing.
If it was a straight up analog emulation though I'd be saying different but I like the idea of these hybrid plugins. not limiting ourselves and giving us different options to play about with
warm up your voice men
Did you not see the notice at the very start? Haha
@@PaulThird 😅
version 2 has now been released. You can use it as a VST within your DAW
And it works 🤓
@@PaulThird can it measure 2 plugins in series so you can see how bypassing 1 plug-in impacts on the other?
You can only do that by loading metaplugin into it