Airwindows ZLowpass: Mac/Windows/Linux AU/VST

[bungle]

Every host I try is as bad as this, I even add OTT after a while to make it very very obvious, the output and input are both way way down, nowhere near supposed unity, and as you can hear, it is horrible.
This is on my laptop at work though and that only runs at 44.1 or 48khz, are these now only usable at 96khz ?

Note this is just a very simple unison synth, nothing else.

EDIT*
Nah just tried it on two other comps at 96khz and in more hosts, exactly the same, I have no idea how nobody else is hearing this.

[gaggle of hermits]

i've kinda replicated the noise with ott but it took some work - unison synth input with high resonance with the gain at odd levels. i don't think this is zipper noise as such i think it's an artefact of the filter itself, maybe some internal clipping or instability, that gets exposed as high-frequency noise.

[Breach The Sky]

So I did some testing with Reason at 44.1. I tried automating freq, as well as cv-modulation. Heard nothing unexpected (?). Added OTT and glitches started happening instantly. And to be honest, I can't hear anything wrong before you add OTT in the clip?

Anyway, I did some more testing, this time I replaced ZLowpass with Reasons built in "Sweeper" filter. Added OTT, nothing weird. But then.. I added True dynamics before OTT and glitches started happening again. At this point, if I even touch the out level in TD, instant glitch-fest ensues. Remove OTT, and problem goes away.

Don't know what to make of this. To me it seems like OTT does something to certain signals. I'm not an avid OTT user, idk. But you had the problem anyway right?

[bungle]

OTT is just bringing out anything that is quite low in the audio via upwards compression, you can hear the zipper noise before I add it too if you listen carefully.
There is no way to automate these filters without zip noise unfortunately, the best thing to do is use multiples on two sends, then just automate the sends, its nothing like a filter sweep, but it is what it is.
Hopefully some coder can pick it up and do what is needed to get it automatable with no zip noise, I know Chris wants to move on, can't expect him to do it.

[jinxtigr]

You might find that the Surge folks do that for you. You'll have to use Surge, and you have to wait until they do it, and I already know the sound will change in other ways too.

But it's all MIT-licensed open source, and I hang out with Surge and know they're thinking of adapting this to their internal filters (that can operate on a note-by-note basis, not as an effect). If they do, they may well fix what you need.

Or… it's an artifact of how these filters literally clip inside the filter algorithm and have additional boosts between each stage of the filter JUST to make them internally clip a little more, something that I know flips out the biquad filters and that I'm doing on purpose to produce a gnarly effect that's weirder than vanilla biquad filters… which would mean anybody doing this would get obvious noises when the biquads are manipulated because biquads DO get weird when you change the coefficients on the fly, and I know that the real sampler's filters are biquads with a hard clip in the middle of the algorithm, known to do strange things.

If you recalculate all the coefficients for every sample… firstly, I'm not convinced that's what the real sampler does, but secondly the biquads will STILL be upset, just in a different way. It's just not a method that lends itself to manipulation. You can try and find somebody to do a version that recalculates all the coefficients every sample. It might not get you a clean sound. I'm told a state variable filter will do that more cleanly… but I'm also told that the Z-Plane stuff is NOT state variable filters, but biquads, from folks who have investigated the actual code (which I haven't).

[jinxtigr]

Also… can you not hypercompress only the extreme highs on a filter that's designed and built to filter OUT the extreme highs from what is expected to be a very harmonically dense signal? There are multiple lowpasses including two in the output stage just to handle what is often a really gritty, clipped output. None of these are meant to let you have super-highs unless you've put the filter there. If they're not set to allow super-highs, there's very little signal up there from any source, and even less from the original audio going into the plugin (even if you're distorting).

Put your OTT BEFORE the ZLowpass. Multiband compressors are evil anyhow. I don't know what you expect from putting a compressed ultra-high band on a 'color' lowpass filter and then running non-bright sounds through it while lowpassing. What is it supposed to be extracting from the signal? If you were distorting hard enough you'd at least get some output-stage grind (which in the Z series filters is very bright and edgy on top indeed) but you're not even distorting.

Not going to fix this one for you, sorry. You should not be compressing and cranking highs that aren't supposed to exist in the audio you're producing. There is nothing for OTT to get, and it's the wrong tool.

…except, of course, when you are using it specifically to do just what it's doing, grab glitchy digital grime out of the noise floor and throw it in people's faces for effect. And that is totally valid, and you should do that. You should be distorting ZLowpass when you do that, though, it will be juicier. And if you're TRYING to get something smooth, you should not be using OTT for that at all.

[bungle]

Woah, blaming multiband compression and OTT, sorry mate, but way off base here, this is a technique used all the time in Reese/Neuro bass design, using a filter for movement and then multiband upward compression to flatten back out the audio for more processing.
An area you specifically said these were for, or did you think EMU filters just distorted beats?

I get it, you have no interest in fixing it, nothing wrong with that, nothing at all, and nobody ever implied there was anything wrong, but don't blame other well known and used tools/plugins, thats bad form.

OTT is built in to one of the most popular synth plugins ever (Serum) with a huge range of filters, non of which have zip noise when OTT is switched on, and no im not dissing your work, Im pointing out that you shouldn't diss others work.

[gaggle of hermits]

no-one is blaming the tools but you.

[crickey13]

don't blame other well known and used tools/plugins

Chris is criticizing multiband comps in general, not just OTT. Instead of using a Z-series filter or any other dirty sounding filter for that matter, it is obvious that you should simply use a clean filter in this context so as to get a clean sound, given that OTT will accentuate the artifacts produced by any character filter.

[jinxtigr]

Woah, blaming multiband compression and OTT, sorry mate, but way off base here, this is a technique used all the time in Reese/Neuro bass design, using a filter for movement and then multiband upward compression to flatten back out the audio for more processing.
An area you specifically said these were for, or did you think EMU filters just distorted beats?

I'm sorry. I was getting testy, because I really do erm… 'dislike' a few classes of processing. One of 'em's FFT based spectral processing, which I feel just always mp3-izes the sound of things whatever else it does, to the point where I've hardly ever been able to like anything about it. The other is multiband dynamics processing. And of course I've long been pretty snippy about Soothe and stuff like it: it more or less does the opposite of what I like in sound.

The reason I get worked up about multiband is, it's a go-to (notably, OTT in particular) that people reach for, always to make stuff LESS of what I like. I don't know of an exception, and it freaks me out that people go for it on general principles. You're not, but it still happens, that people do this in order to fight against sounds that I strongly prefer… in context. Cos it's always in context, whether people know it or not.

I apologize, I got carried away. Thing is, I've got stuff that's smoothed and stuff that's not. PurestGain and especially PurestFade are specifically smoothed against zipper noise. The Z series filters? On the one hand I don't want to recalculate the filters every sample (or even double up the processing and interpolate along the sample buffer, which is what I think the Emu version DOES actually do- or it might, because that's what it would take to do the morphing stuff Z-plane does). On the other hand, what if someone wanted to apply a squarewave modulation at the rate of one of the bass notes? I want it breakable. I want it glitchable, like my Glitch Shifter plugin: and specifically if you automate a sudden snap to another setting I want it to do that in one instant, and not try to smooth to it… because with biquads, smoothing to the new value (unlike with a gain trim) will definitely NOT give you a clean sweep. It can give you a voltage spike 40 dB over the level of the signal.

That's not true with the Z filters because they internally distort, and they have an output stage, so there will be no 40 dB spikes from trying to apply interpolated coefficients to the biquads. The glitches from whatever source are going to be both clipped to maximum output, and also slightly softened as if they're going through a transistor output stage on hardware, because that's the double-filtering on the output stage, similar to but not exactly the same as the one in Mackity. That's why when you distort wildly and sweep up you can get a volume jump similar to the one the real hardware unit does…

I'm sorry for jumping all over you. I do know I'm not the guy to bring you the evolution of what these are, but they ARE MIT licensed, meaning you or anybody else could take them and continue (they're quite simple really, though what you want out of them would make them at least four times more complicated to execute). The problem we've got, and I'll try harder to not get snippy about it, is that you're wanting a lot of work to change the plugins and make them worse in several ways by my perspective. So I can tell you anything you'd need to know about where they're at, I can probably even speculate and advise on what you'd have to do to get where you want to be with the code (though I could be wrong, or not able to implement it properly), but I can't do it with my versions of the plugins. I need them to stay with the balance of tradeoffs and choices I had for them

[jinxtigr]

Also, confirmed: if you do see a take on this in Surge, it will by default be fully and correctly smoothed. They make different choices in Surge-land, so that might be your chance… and once done, if you needed the same thing but as a standalone filter, anyone could begin from what Surge did, and do standalone filters back again, so long as they stuck with Surge's GPL licensing

[rafa1981]

You can recalculate the filters at a fixed rate when smoothing. It doesn't need to happen sample-wise.

I would recommend you to use C++ classes, #include directives etc on your codebase, so you don't have to copy paste your dithering stuff, Biquads code, etc on every plugin for every platform.

[bungle]

no-one is blaming the tools but you.

Erm nope.
Never blamed any tools, pointed out zipper noise, nothing more.

don't blame other well known and used tools/plugins

Chris is criticizing multiband comps in general, not just OTT. Instead of using a Z-series filter or any other dirty sounding filter for that matter, it is obvious that you should simply use a clean filter in this context so as to get a clean sound, given that OTT will accentuate the artifacts produced by any character filter.

Erm nope, these are not smoothed for zipper noise, I like the character, I want the character, I pointed out zipper noise, not character, huge huge huge difference.

Stuff

Thank you Chris, I know you hate OTT, you have made that clear in the past, in this instance it is used as a flattening sound design tool though as you know, I think you may have missed that, and also the fact that I did say you are moving on and hopefully somebody else can do what needs to be done to get these zip free (Not you, I know, you have other things to do) and my posts were more about not understanding how nobody else was even hearing the zip noise, but I am guessing that most here don't do DnB bass design with heavy movement automation and rework.

Thank you for these by the way, as I posted earlier, even with the zip noise, I can super boost the zips and use them specifically for Spore type glitch bass anyway.

(FYI I know somebody else that will be implementing these in a synth and smoothing them, they already have Mackity as an effect in the Alphas on my recommendation, so thanks)

Oh FYI, I can get the notch to automate with no zip noise, there is a very specific pole setting that allows it

[jinxtigr]

Oh FYI, I can get the notch to automate with no zip noise, there is a very specific pole setting that allows it

Cool! happy ending.

Or, rather… remember Srsly? And how, not right away but later, there was a Srsly2 which was the same thing but taken to a whole other level with a factor that wasn't part of the first Srsly, and it was the second Srsly2 that served as more of a functional replacement for using a Hughes SRS box the way folks were using them?

It's like that. I have to move on for now. If I return to the Z filters it won't be to replace them, it'll be because I put in some work on coefficient interpolation in sample blocks and mastered (or came up with my own way of doing) that technique which isn't present in the current Z series filters, and then it would be another four weeks or so adapting whatever I learned to each existing Z filter, while leaving the originals there (like how there's still a Srsly) for their own purposes, which do exist.

If you want to call that the only REAL happy ending, that's okay with me, and you'll have to wait and I don't know how long. I've got a happy dumb filter in the works that a bunch of people will really enjoy, on the concept of 'tubey fatness' and one-knob make-betterizers (which are normally not my jam). There might be other things I need to do, also.

But for now, I'm happy you found a limited use for my days in Z-land, and that there's a ZNotch setting that works. I assume you don't mean Poles of 0.0

[bungle]

Yeah, I didn’t actually say limited use either
Glitch bass is a huge part of DnB sound design, the zip noises make this a simple thing to set up.
Never ever equate me stating a fact like “This has zipper noise” as “This is sh!t” not even close, if I thought that, I wouldn’t post at all.
Maybe you are not getting told enough, but right now you are the most respected developer in VST land in all the circles I move in