The No.1 Website for Pro Audio
Lets do it: The Ultimate Plugin Analysis Thread
Old 9th September 2020 | Show parent
  #3661
Lives for gear
 

Quote:
Originally Posted by Jeffguitar View Post
Not to take anything away?. Sounds like you are attempting to take something away for sure, his skill level. I don't hear harshness, how about giving us an example of harshness in a couple of his mixes and let's just admit that you could say that because as you know this is just an opinion based thing and others could listen to the same mix and say it doesn't sound harsh, actually, they could say it sounds dark with too much mud area while others could say it sounds perfect.
Sigh.. no, I really do not want to take anything away from Warren. I DO think many of his examples over on youtube sound quite harsh. I've also been a part of PLAP as I was there to learn tips and tricks like anybody else.. so yeah, I've heard many of the mixes.

Some examples, in case you have them available:

Alaina Blair - Love It or Leave It
Alexx Calise - Wasting Away
London Lawhon - Locked Up (this one in particular is most definitely harsh in the chorus)
We Were Astronauts - Keep It Together (definitely harsh esses and some other stuff in that 8 to 10k area).

Yes, overall I do find his mixes to be leaning quite hard in the upper mid range and highs. Do I still think his overall balance and mixing skills are good? Yes, definitely.

I'm also absolutely sure many people would find the same mixes I think are quite harsh to be perfectly fine. There's nothing new there. We all have our preferences.

I'm sorry you feel like you have to personally attack me and make assumptions about my motives for my opinion. I AM being honest here. There is no ulterior motive.. and yes I DO think Mr Huart's mixes could sound a bit better if he took special care with plugins that cause aliasing. This applies to all of us using plugins.

Quote:
Originally Posted by Jeffguitar View Post
Quote:
Originally Posted by bmanic;
, the lower the basic fundamental of a signal is, the less problems you have with aliasing. You can quite safely use a lot of processing on a smooth bass for instance. Problems can arise when the bass is picked or slapped. Then you immediately have that clicky 1.5 to 3kHz which _will_ produce aliasing if pushed hard into saturation plugins without oversampling.
As I'm sure you know, Warren, like many, maybe most of us will take his DI bass take, high pass it hard core and load quite a bit of saturation on it while low passing the amp track. So the saturation he uses is fairly extreme, and he uses the built in saturation in PT.
Indeed. A perfect place where one would benefit a lot from taking care of the chain.

Quote:
Originally Posted by Jeffguitar View Post
Quote:
Originally Posted by bmanic;
I know you guys want to go for the straw man argument of ".. but other people have done perfectly fine mixes without any aliasing issues!". Yes, but this doesn't negate the problem with aliasing itself. It just shows the skill of the person being able to do good mixes with the tools they choose. They use their ears and choose their tools and the amount of "abuse" accordingly.
And you guys want to cram 6 instances of the same comp in series and claim this is dangerous for us when NORMAL MIXERS DO NORMAL MIXING.
Way to go overboard with the exaggeration. Nice one.


Quote:
Originally Posted by Jeffguitar View Post
For instance, as I said, I have X2 set in my Cakewalk Bandlab. I have DMG Track comp, I usually use a couple of them and set at x4, which makes x8 with Bandlab at X2. I use R comps somewhat, I use one instance of Iron if I want. I use MU at x4 on render.
The point is, people on your side go just as far past what normal mixers do as well, easily as much as our side go in the opposite direction. What most all want to know is, IS ALIASSING A PROBLEM FOR NORMAL MIXERS MIXING, YOU KNOW, NORMALLY. Like using some plugs that DO over-sample, some that don't, so that don't but are not emu designs like the Rcomp. Normal mixers do a REASONABLE mix of various tools.
Why on earth are you talking about "sides"? There are no sides here. This is not US politics.. feels to me like you are trying to create a conflict when there is none.

Everybody is of course entitled to do whatever they please. My only point here is to show that aliasing is not a myth and not something you should simply ignore. It can have real consequences and cause unnecessary inharmonic high frequency content when having multiple non-linear plugins in series. That's all.


Quote:
Originally Posted by Jeffguitar View Post
So will alliasing be a problem in I have a 16 track song, I use 1 Iron mix bus, say, 5 track comps at x4 or x8, 2 Decapitators, 3 Rcomps, Limitless limiter, .
You decide if it is a problem or not. If you have all of those plugins in series, then yes.. you would definitely hear a big difference if you rendered the song at a higher sample rate and then down sampled.

If they are in parallel, each effect on it's own track, then no.. it should be fine, though I suspect some people would hear benefits still, mainly due to Decapitator and SPL Iron.


Quote:
Originally Posted by Jeffguitar View Post
And 1 last question. Have you not heard some say that x8 os can also sound bad to them, that it can change the sound in an unpleasant way?
Like anything else in DSP land, there is no free lunch. Yes, up and down sampling can cause audible artifacts too. They are different to aliasing though. To my ears up/down sampling usually messes with the transients slightly. They get slightly "harder" and more "clicky". And like with any DSP solution, some up/down implementations are better done than others.

But is it worse than heavy aliasing? No, not in my opinion. I'd even take a relatively poor up/down sampling implementation instead of heavy aliasing. Especially if it is in a saturation plugin that I want to use a lot.

It's all about context and what works where. I can usually fight clicky transients but it is, as far as I know it, impossible to remove aliasing once it's there.

My general rule of thumb for minimizing aliasing is this: If you have several heavily saturating things in series, one after another, then take special care in which plugin you choose.. or render that track/part at a much higher sample rate and then down sample it.

If I have multiple busses and complex routing with many plugins, which ends up going in series, then I do the same thing. Either up and down sample before render, or choose plugins that do internal oversampling well.
Old 9th September 2020 | Show parent
  #3662
Lives for gear
 

Quote:
Originally Posted by Nokturnal View Post
Looks like we are going through the aliasing chapter again… here’s Dan Worrall’s take on it:

It's a very good video, showing quite clearly some of the problems with aliasing.

Unfortunately Dan was apparently not aware of the TDR Ultrasonic Filter plugin when making that video so some of his points against high sample rates are umm.. distorted.

But yes, this video shows clearly that high levels of oversampling is very effective in combating aliasing. Especially with stuff in series.
Old 9th September 2020 | Show parent
  #3663
Quote:
Originally Posted by bmanic View Post
Unfortunately Dan was apparently not aware of the TDR Ultrasonic Filter plugin when making that video
I'll try TDR Ultrasonic Filter, although still and probably will always be in alpha stage.

Is this the only plugin that can do this "trick"?
Are there any alternatives?
Old 9th September 2020 | Show parent
  #3665
Gear Addict
Quote:
Originally Posted by bmanic View Post
It's a very good video, showing quite clearly some of the problems with aliasing.

Unfortunately Dan was apparently not aware of the TDR Ultrasonic Filter plugin when making that video so some of his points against high sample rates are umm.. distorted.

But yes, this video shows clearly that high levels of oversampling is very effective in combating aliasing. Especially with stuff in series.
Thank you very much for your post! How did I not know about this?

I was finishing up an acoustic mix and placing US filter after each instance of nonlinear processing which does not have OS really cleared up the mids, imaging and gave better front to back depth. And I was not using much nonlinear processing to begin with.

It is crazy how much unwanted stuff gets generated and muddy up mids.

Are there any downsides to the TDR filter that you have found?

Couple of past days I have had some great revelations because of fine folks like you sharing information here in gslutz.
Old 9th September 2020
  #3666
Lives for gear
 

Question.

Why no Filter for 44.1 samplerate?
Old 9th September 2020 | Show parent
  #3667
Gear Addict
Quote:
Originally Posted by RJHollins View Post
Question.

Why no Filter for 44.1 samplerate?

My understanding
is that ultrasonic filter is needed because the generated "stuff" that is in range from 20k-44/48k(nyquist freq) is doing the inter-modulation harm. It is not heard and not needed but bounces back when further processing is applied.

At base-band you do not have any ultrasonic frequencies to filter.
Old 9th September 2020 | Show parent
  #3668
Lives for gear
 

Quote:
Originally Posted by Stikkers View Post
I'll try TDR Ultrasonic Filter, although still and probably will always be in alpha stage.

Is this the only plugin that can do this "trick"?
Are there any alternatives?
Any EQ plugin that allows you to go all the way up past 22.05kHz when at high sample rates will do the trick. Just put a lowpass filter there above it.
Old 9th September 2020 | Show parent
  #3669
Lives for gear
 

Quote:
Originally Posted by kuulart View Post

My understanding
is that ultrasonic filter is needed because the generated "stuff" that is in range from 20k-44/48k(nyquist freq) is doing the inter-modulation harm. It is not heard and not needed but bounces back when further processing is applied.

At base-band you do not have any ultrasonic frequencies to filter.
Yes. It's probably something like that.

I actually have no clue on how it works. I just trust Fabien's genius and plonk down the plugin according to his instructions.
Old 10th September 2020 | Show parent
  #3670
Gear Nut
 

Quote:
Originally Posted by chrisj View Post
Um… we can agree to disagree on truncation error: as someone who dithers to the floating point bus on every plugin, clearly I have no interest in restricting myself to only what's correct 'within the audible range'. And that's just fine, I'll continue to 'overdesign' around wordlength and you can't stop me

Aliasing is a completely different kettle of fish. You can make it as loud as you like. You can make it substantially louder to the human ear than the source audio: just apply some plugin with a sharp discontinuity in the transfer function, and then send in your hi-hats or supersonic test tones. I mean, it'll be impossible to observe aliasing on earthquake noises, but we're dealing with music and we know some of the instruments that we'll be hearing. The sibilance of a human voice with NO instrument can produce stuff that will interact real poorly with aliasing.

I deal with aliasing in my stuff by basically doing my distortion in the analog domain since I don't like the sound of heavily overprocessed digital audio I'm kinda stuck just avoiding sharp discontinuities in the transfer function, whether that's saturation algorithms, compression, what have you. My most successful work in that direction is all about avoiding the conditions that cause aliasing in the first place. And I'll be the first to admit that I make stuff where you can make it alias good and hard if you try hard enough. My hope is that it balances out with the other choices I make, and that people aren't so compelled to distort that hard in the digital domain.

I think it sounds bad, and it can be incredibly obtrusive and audible, and by definition it moves energy from the super-highs (harder to hear) to smack in the middle of the audible range where we can hear problems the easiest. It all depends on what kinds of sounds you're dealing with and how much margin for error you've got in terms of sample rate. I made a Eurorack module go from extreme aliasing to very little, simply by changing the sampling rate of the Teensy from 44.1k to 340k (it's a really naive algorithm so it proved feasible to just overclock the primitive DAC)
Hey man I respect all you said here.

I'm no dsp expert, i'm a producer/mix engineer...

I just read what some of you developers have been saying in this thread to educate myself a little bit, and my conclusion was that this isn't stuff I have to be worried about.

Specifically the spl iron truncation error stuff.

I'm very grateful for the insight, but I doubt I could make this audible even if I tried! Based on what you guys are saying on pg 114 on onwards.
Old 10th September 2020 | Show parent
  #3671
Gear Nut
 

Quote:
Originally Posted by strange loop View Post
Out of curiosity for what they had to say I searched the thread but could not find these posts.

Regardless, it's up to the individual to decide whether aliasing matters to them. This goes both for those who think it's not a problem and those who prefer to avoid it.
If your interested, pg 114, starts around post #3416 , and goes on for a few pages x
Old 10th September 2020
  #3672
Gear Addict
 

@ smellypants Thanks. I remember that discussion. Not all the details but I was involved and it was interesting.

Reading through it again doesn't change my opinion on aliasing.

The focus early on was truncation. I see in one of my own posts I tested for truncation in the Maag EQ4 (which had been reported, as I recall, to also truncate the same way that the Magnum K and SPL Iron are claimed to do. I was never sure if they actually do truncate.) And my own tests with EQ4 showed any added noise (whether from truncation or whatever) from the plugin was more than -130dB below the signal. Andy (Cytomic) reiterated that noise more than -120dB below the signal can be effectively ignored, and even up to -100dB below is generally fine. My conclusion was "maybe it does truncate but not at levels that matter." I didn't test SPL Iron for that but it may well show the same low levels and so likely nothing to worry about. And it sounded like most of the developers agreed this was too low level to care (and unlike aliasing produced via sequential nonlinear processing I expect truncation is unlikely to stack up the same way so I remain unconcerned.)

But when the discussion turned to aliasing I don't see anything from experienced devs saying "don't worry about it, you'll never hear it because it's always too low in level." And the main developers contributing to this thread are ones who put extra effort in to their plugins to minimise aliasing. Why would they do that if they thought it didn't matter?
Old 16th September 2020 | Show parent
  #3673
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by bmanic View Post
Any EQ plugin that allows you to go all the way up past 22.05kHz when at high sample rates will do the trick. Just put a lowpass filter there above it.
I doubt many EQ plugins have a 100dB/octave slope tho...
That TDR filter is not a typical lowpass filter. It has a pretty sharp cutoff.
Old 17th September 2020 | Show parent
  #3674
Quote:
Originally Posted by monomer View Post
I doubt many EQ plugins have a 100dB/octave slope tho...
That TDR filter is not a typical lowpass filter. It has a pretty sharp cutoff.
True! And it's a lot lighter on the CPU too. So, i'll just be using that ultrasonic filter thing.
Old 2 weeks ago | Show parent
  #3675
Lives for gear
Has anyone put Softube Harmonics through the scope yet. Is the the saturation dynamic and legit, and does it alias?
Old 1 week ago
  #3676
What I would like is an oversampling plug-in that operates much like MeterPlugs Perception or LetiMix GainMatch: where you have a plug-in on the front-end of a chain and another on the back-end. Everything in between would be processed at 2x, 4x, 8x, or 16x oversampling, then filtered back down to native.

I know we already have DDMF Metaplugin to do oversampling of plug-ins at lower sessions rates... but I find the interface messy and disruptive to workflow. I prefer the organized chain and routing flexibility of Bluecat Patchwork; but there is no oversampling option.

The biggest drawback to the Metaplugin/Patchwork approach—for me, anyway—is that I don't really care for shells that contain multiple plug-ins, like Slate VMR or Patchwork... I like to glance at a channel and see immediately what plug-ins are being used and go to them directly. I don't like the additional step of opening a shell to access (and remember) what's in it and again opening a plug-in to tweak it.

I work mostly at—and am also given—48k sessions and don't want the overhead and issues of 96k or 192k... however, aliasing and proper filtering still need to be addressed for a lot of plug-ins that don't do well at 48k. But what we need is a more elegant and flexible solution. AAX, AU, VST

I'd love to get the thoughts of some developers and engineers who are more educated on the minutia of all this more than I am.

In effect, I'd love to take a few really important chains in a 48k session—like a lead vocal—and put a plugin on the front end (a la Perception or MatchGain) that chooses the OS factor (2x, 4x, 8x, etc.); any plug-ins that follow would be processed at this higher rate, (a la Metaplugin); then a final plug-in would handle the downsampling and ultrasonic filtering back to the native session rate.

Thoughts? I this doable?
Old 1 week ago | Show parent
  #3677
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by robshrock View Post
The biggest drawback to the Metaplugin/Patchwork approach—for me, anyway—is that I don't really care for shells that contain multiple plug-ins, like Slate VMR or Patchwork... I like to glance at a channel and see immediately what plug-ins are being used and go to them directly. I don't like the additional step of opening a shell to access (and remember) what's in it and again opening a plug-in to tweak it.

Thoughts? I this doable?
I don't think this is possible. The DAW will always see every individual plugin as working at the system sample rate. You can't simply temporarily up the sample rate of your entire channel strip.
So the only sensible solution is to host the upsampled plugins inside a host plugin. On the outside the plugin looks like a normal plugin to the DAW and on the inside it hosts other plugins at the higher rate.
Old 1 week ago | Show parent
  #3678
Quote:
Originally Posted by monomer View Post
I don't think this is possible. The DAW will always see every individual plugin as working at the system sample rate. You can't simply temporarily up the sample rate of your entire channel strip.
So the only sensible solution is to host the upsampled plugins inside a host plugin. On the outside the plugin looks like a normal plugin to the DAW and on the inside it hosts other plugins at the higher rate.
Makes sense. I guess level matching is drastically more simple a process than oversampling. lol

That said, I do wish there was a simpler/cleaner interface for Metaplugin or something else like it. It feels like working in the Logic Environment... and it doesn't feel comfortable to me for fast, daily workflow.

Thx
Old 1 week ago | Show parent
  #3679
Gear Nut
 

I can confirm that. This would only be something that can be done at the DAW level... a plugin in a chain has no access whatsoever to the plugins that come before or after it, and those plugins have to be notified that they are supposed to operate at a higher sample rate, it's not sufficient to just send them twice the amount of audio data.
What I could imagine is an "oversampling bridge" similar to what I've already released as a 32-64-bit bridge: a solution that would create oversampled versions of any plugin and store them in your plugins folder. So along with MyFavEQ.dll you'd have MyFavEQ_2x.dll and/or MyFavEQ_4x.dll (insert your favorite plugin file suffix instead of dll).
Old 1 week ago | Show parent
  #3680
Quote:
Originally Posted by robshrock View Post
Makes sense. I guess level matching is drastically more simple a process than oversampling. lol

That said, I do wish there was a simpler/cleaner interface for Metaplugin or something else like it. It feels like working in the Logic Environment... and it doesn't feel comfortable to me for fast, daily workflow.

Thx
Maybe ask DDMF to make a metaplugin light? With only the oversampling option.
Old 1 week ago | Show parent
  #3681
Lives for gear
 

Quote:
Originally Posted by Stikkers View Post
Maybe ask DDMF to make a metaplugin light? With only the oversampling option.
.... and maybe for just a single plugin enclosed.

Old 1 week ago | Show parent
  #3682
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by robshrock View Post
Makes sense. I guess level matching is drastically more simple a process than oversampling. lol
Level matching/changing doesn't mess with the internal sample format (rate and bit depth) of the daw. It works on any sample format. That's why you can have those.

Quote:
That said, I do wish there was a simpler/cleaner interface for Metaplugin or something else like it. It feels like working in the Logic Environment... and it doesn't feel comfortable to me for fast, daily workflow.
Well, technically DAW's could facilitate in hosting such a plugin, but it would take some effort to set up a standard and a lot could go wrong. There would also be no real market for such upsampling plugins. They all would do basically the same thing so no real advantage for one over the other. And in the end it would probably be more efficient to actually have the DAW upsample a whole channel or simply run the entire engine at a higher rate.
I think that having the DAW do it for you (maybe with a button on each channel) would probably be the best way for such a function to exist.
Old 1 week ago
  #3683
Gear Addict
 

You don't want the whole chain oversampled anyway. If you're doing that on any channel which has at least one nonlinear plugin chances are you're doing it on every channel, or at least most. May as well run the entire session at the higher rate.

What I want is for you to be able to tell the DAW to oversample selected plugins. That way I can take the few plugins I have which don't already have internal oversampling and which can benefit from oversampling and select only them. If I can save the oversampling option as a default for that device (Ableton allows this kind of thing) even better.
Old 1 week ago | Show parent
  #3684
Quote:
Originally Posted by Stikkers View Post
Maybe ask DDMF to make a metaplugin light? With only the oversampling option.
Not a bad idea... as I said earlier, I would love it to feel more like Bluecat Patchwork if we must have a shell matrix paradigm.

I’m hoping some of the more clever designers could chime in. Help us, please.
Old 1 week ago | Show parent
  #3685
Maybe AnalogObsession can help us build a container plugin which can oversample the plugin you load into it. The oversampling algorithm that Analog Obsession plugins have implemented is very good!

How can you reference a user so that he will see this post?

Just sent him a private message.
Old 1 week ago | Show parent
  #3686
Gear Maniac
 
Anthony Quinn's Avatar
 

Quote:
Originally Posted by Stikkers View Post
Maybe AnalogObsession can help us build a container plugin which can oversample the plugin you load into it. The oversampling algorithm that Analog Obsession plugins have implemented is very good!

How can you reference a user so that he will see this post?

Just sent him a private message.
I’ve been avoiding Analog Obsession because I couldn’t tell if he was doing oversampling, so didn’t take the time to install or test. This is going to be a good day.
Old 1 week ago | Show parent
  #3687
Lives for gear
 
zvukofor's Avatar
Quote:
Originally Posted by Anthony Quinn View Post
I’ve been avoiding Analog Obsession because I couldn’t tell if he was doing oversampling, so didn’t take the time to install or test. This is going to be a good day.
The only AO plugin that does not alias is NSCAR.
Old 1 week ago | Show parent
  #3688
Quote:
Originally Posted by zvukofor View Post
The only AO plugin that does not alias is NSCAR.
Not true! Check STEQ.
Old 1 week ago | Show parent
  #3689
Gear Maniac
 
Anthony Quinn's Avatar
 

So just those two? That would save me some time and possible disappointment, a lot of plugins over there.
Old 1 week ago | Show parent
  #3690
Quote:
Originally Posted by Anthony Quinn View Post
So just those two? That would save me some time and possible disappointment, a lot of plugins over there.
I've only tried three, and only tested one. So, can't answer your question.
📝 Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 272 views: 70610
Avatar for Roberto Alvarez
Roberto Alvarez 4th May 2017
replies: 80 views: 29638
Avatar for projektk
projektk 30th January 2012
replies: 530 views: 72036
Avatar for joe_04_04
joe_04_04 12th August 2015
replies: 453 views: 70490
Avatar for djrustycans
djrustycans 15th October 2020
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
🖨️ Show Printable Version
✉️ Email this Page
🔍 Search thread
🎙️ View mentioned gear
Forum Jump
Forum Jump