BuzzCut (The Erosion Clipper)

[EbN]

AUTO-ADMIN: Non-MP3, WAV, OGG, SoundCloud, YouTube, Vimeo, Twitter and Facebook links in this post have been protected automatically. Once the member reaches 5 posts the links will function as normal.

Anybody tried this new clipper? Supposedly using a new way of masking the distortion/aliasing with noise, kinda like dithering is masking truncation..

https://lusidmusicuk.com/products/buzzc ... on-clipper (https://lusidmusicuk.com/products/buzzcut-the-erosion-clipper)

There are probably better ears and judges here than me, so curious what more experienced people think of this

[BCollins]

Downloaded today and compared it to a vast amount of clippers I own on 2 bus.

This clipper is pretty unique and sounds incredibly open and clean. Very true to the source.
Regarding the dithering / noise approach, Airwindows and Maat Fidef come to mind .

I highly recommend to do your own testing - might be the one you´ve always been looking for!

[pekbro]

It's pretty cool and worth a look imo
(seriously)

[jinxtigr]

Downloaded today and compared it to a vast amount of clippers I own on 2 bus.

This clipper is pretty unique and sounds incredibly open and clean. Very true to the source.
Regarding the dithering / noise approach, Airwindows and Maat Fidef come to mind .

He's not using anything of mine, though I'm talking privately with the guy in efforts to help him get the mono AU to happen. I hope he does, because if he's able to make a mono AU version it'll also be N to N and work in surround, if he does it right. He won't have to be using my code to do it, either, although in another forum he's got a guy asking him for 'slew dithering' (which is not a thing) but it came up relative to my Chrome Oxide which is noise applied only to the slew, and was from somewhere between 2007 and 2011.

And also asking for floating point dithering: far as I know I am the ONLY one doing that, so don't expect it from him unless he's all in on the MIT-license thing

Personally, I think it's a great idea. As he has it, it's most suited for how he's using it, in heavy bass-music that must be extremely loudenated. I've already said I'm not going to do a version of his plugin so if you want that you'll have to get it from him (or, presumably, a vibe-code imitator, but I see a lot of those guys will get tangled up in a sea of requests and flounder)

What I did get out of seeing his post (haven't tried the plugin) is a renewed appreciation for randomizing clip onsets and departures. I've had ClipOnly2 out for years, doing its thing and being a very good 'not edgy' normal-intensity clipper. It uses a fixed proportion of how much to soften, and I've assumed that diffracts aliasing a little bit. However, nothing is stopping me from noising up just those two clip samples, and while it won't do the 'Erosion Clipper' thing nor do I need to do that, I bet I can use the technique way more subtly to further enhance ClipOnly. I should be playing with that next week, and if it works, it'll stop repeated clipping waveforms from aliasing, though probably without throwing quite as much noise, and without the 'dithering the clip threshold means you can track bass beyond full clip a little, it'll act like a softclip except it's a blast of noise on bass peaks'. I won't have any of that but for my purposes I won't need it

All in all a very cool day to see this come out! I wish him the best and hope my advice helps with the mono thing. Back in the day (what, twenty years ago?) I got help from Sophia Poirier of DestroyFX, and Zaphod of Acustica. In fact I think I got the help from Zaphod, on KVR

[Zeisner]

Ask yourself: Why does it feature oversampling if this "stochastic erosion" would reduce aliasing? And what's the point of adding even more noise? Sure, as a special effect it can be useful for genres like industrial. But as a general tool?

[LusiD_Music_UK]

Ask yourself: Why does it feature oversampling if this "stochastic erosion" would reduce aliasing? And what's the point of adding even more noise? Sure, as a special effect it can be useful for genres like industrial. But as a general tool?

Excellent question and I appreciate the opportunity to push back. What BuzzCut offers is “choice” - do you want “foldback aliasing” or “noise”. Just like dithering doesn’t “solve” truncation distortion, it replaces it. Likewise erosion clipping doesn’t “solve” foldback aliasing. It replaces it. Most people agree that random noise sounds better than digital artefacts. This is why dithering is basically universally accepted as superior. Oversampling is there again for “choice”. If you use the 16x oversampling option then there is a lot less aliasing to replace. You’re absolutely free to not use erosion clipping or oversampling, it can act like a normal hard clipper or soft clipper too. I do genuinely understand the skepticism.

[Zeisner]

What BuzzCut offers is “choice” - do you want “foldback aliasing” or “noise”.

I quote from your webpage:

By modulating the threshold with shaped noise that shatters the source of inharmonic foldback aliasing before it happens.

That is nonsense. Without oversampling (and filtering the frequencies above the Nyquist frequency) there is still aliasing. By adding noise above Nyquist you only make that noise alias as well. You can "mask" aliasing without oversampling by simply adding noise filtered with a highpass filter. But does it sound good? For certain genres that require noise, sure. But as a general tool? Forget it.

Just like dithering doesn’t “solve” truncation distortion, it replaces it.

The solution to quantization noise is not dithering, it's higher bit depths, the equivalent to higher samplerates.

Likewise erosion clipping doesn’t “solve” foldback aliasing. It replaces it.

No, you're simply adding noise to preexisting noise.

Most people agree that random noise sounds better than digital artefacts.

But what sounds best? No noise at all.

This is why dithering is basically universally accepted as superior.

Dude. Streaming services offer more and more ofzen 24-bit streams. Because the quantization noise of such a bit depth falls below the audibility threshold of humans which means no more dithering is required.

[Zeisner]

If I made that plugin I would not have tried to fool customers. Instead I would be upfront, saying what the plugin really does. It can be useful as a tool/special effect hybrid for certain genres. But that has nothing to do with aliasing. Sell it as "collapsing clipper" or something. You would attract the right kind of customers and have a better reputation.

[Andreya_Autumn]

If I made that plugin I would not have tried to fool customers. Instead I would be upfront, saying what the plugin really does. It can be useful as a tool/special effect hybrid for certain genres. But that has nothing to do with aliasing. Sell it as "collapsing clipper" or something. You would attract the right kind of customers and have a better reputation.

So, I'm not affiliated with the dev in any way nor have I tried the product, but I understand enough DSP to be able to tell you that his idea is in fact sound.

Long story short, to reason about why harmonics fold back into the audible spectrum it's useful to note how the timing inaccuracies caused by the sampling are *periodic*.

I find the easist mental model is: imagine a digital square wave, and let's ignore for now the (rare) case where the square runs at an exact division of the sample rate. Each change from high to low should theoretically happen somewhere between samples, right? But if we do it naively, the best we get is high on one sample, low on the next. If the exact time when the transition shoulda happened happens 5% of the way between two samples, next time it might be 10% of the way, then 15% etc etc.

This is a somewhat simplified model. But it's close enough to make the point that the *periodicity* of the timing error gives you an image* of the problem. The timing error *cycles* (at a rate proportional to the diff between the square frequency and nyquist). When soft clipping a signal, the image of the problem is less obvious. But similar dynamics play out in that case also.

Point is, if the process that generates the harmonics is randomized in an appropriate way, the periodicity which caused the folded-back harmonics goes away! The method does in fact replace aliasing with noise, not just mask it.

It is of course true that aliasing and quantization noise can both be solved with higher sample rates or bit depths. At least if by solved we mean rendered entirely unproblematic for human listeners. A truly nitpicky player of the "who is right" game would contend that those things don't solve the problems at all, and they would be strictly correct but also a bit of a jerk. Your point stands that removing aliasing without replacing it with any noise is in fact possible.

But I also don't see the dev claiming otherwise! So to say they're "trying to fool customers" is thus pretty harsh in my opinion.

* Notice I didn't say it's the cause. Because in it isn't. Subtle distinction, but important.

[Andreya_Autumn]

EDIT: Oops, I keep pressing quote instead of edit.

[Bohmes DSP]

It is of course true that aliasing and quantization noise can both be solved with higher sample rates or bit depths. At least if by solved we mean rendered entirely unproblematic for human listeners. A truly nitpicky player of the "who is right" game would contend that those things don't solve the problems at all, and they would be strictly correct but also a bit of a jerk. Your point stands that removing aliasing without replacing it with any noise is in fact possible.

I was so close to being that guy but I talked myself out of it at the last second. Came, went back and found you'd just wrote this

I have a question though:
"Point is, if the process that generates the harmonics is randomized in an appropriate way, the periodicity which caused the folded-back harmonics goes away! The method does in fact replace aliasing with noise, not just mask it."

Why do you get aliasing on cymbal hits or a crowd clapping? There are loads of none periodic sounds that create aliasing.

[Zeisner]

There is only one way to "erode" aliasing (similar to modulating quantization noise instead of masking it) and that would be (pseudorandomly) changing the samplerate per sample, ideally as parallel processes to distribute aliasing artifacts even more across the frequency spectrum. But that still doesn't solve the aliasing issue itself, only oversampling and filtering out all frequencies that would fold back does.

[mcbpete]

(pseudorandomly) changing the samplerate per sample

Curious as to what that might look/sound like - would this essentially be jitter in the frequency (time?) domain, vs dither in the amplitude domain

[Andreya_Autumn]

There is only one way to "erode" aliasing (similar to modulating quantization noise instead of masking it) and that would be (pseudorandomly) changing the samplerate per sample, ideally as parallel processes to distribute aliasing artifacts even more across the frequency spectrum. But that still doesn't solve the aliasing issue itself, only oversampling and filtering out all frequencies that would fold back does.

Again, even that strictly does not! But can indeed render it inaudible. But that's the last time I'm gonna walk that jerk-line, I promise! We agree that silencing aliasing without the noisy side effect is possible. My only disagreement was with the perception that the dev has claimed otherwise!

You might not be interested in a process that replaces folded-back harmonics with noise. Which is fair! In most situations I ain't either. All I'm saying is the process as described can in fact do what was claimed.

"Eroding" aliasing is not a term that has a well-defined meaning in DSP research anyway. It's just the term the dev chooses to use for their product. You seem to feel it sounds misleading. Fair enough, I ain't the judge of that. To me the english word erosion brings about images like... say a waterfall wearing down on rocks over a thousand years. Feels kinda fitting to be honest. But this is obviously subjective. Again, don't wanna play who's right.


Anyway. Again, the folded-back harmonics are found in the relationship between the nyquist freq and the non-bandlimited process. Adding noise to that relationship could indeed be achieved the way you describe, by randomizing the nyquist frequency. But since the thing at the heart of the issue is the relationship, we can just as well change the other factor, the non-bandlimited process. Which seems way easier... and more theoretically "clean" (since you aren't introducing another resampling stage with interpolation etc).

People have done this before in synthesis. Very slightly offsetting the rate of a square or saw each cycle can indeed give you a wave with none of the usual aliasing, but a buncha noise instead. Whether that's a better sound or not is of course subjective. Again, just saying it does work as described!

[Andreya_Autumn]

(pseudorandomly) changing the samplerate per sample

Curious as to what that might look/sound like - would this essentially be jitter in the frequency (time?) domain, vs dither in the amplitude domain

Yup. There is a (seldom mentioned) name for this strategy which presently escapes me... but which I *think* contains the word jitter. It came up in the DSP forum a while ago. Or more likely in an old thread that was necrod a while ago. Maybe the supersaw one, don't remember.

But yeah it would make everything sound noisy! Which if you are making say a tape emulation (or a creative distortion which is what BuzzCut looks like to me I guess) might be just what you want anyway!

But you wanted a super clean high-fidelity result, then the strategy isn't for you. That's the truth in what Zeisner wrote (If I'm reading you right). Regular anti-aliasing measures will work better then.