Airwindows

BitDualPan

TL;DW: BitDualPan is dual pan that’s almost BitShiftGain.

BitDualPan.zip (518k) standalone(AU, VST2)
BitDualPan in Airwindows Consolidated under ‘Utility’ (CLAP, AU, VST3, LV2)

This is by request, and also a sort of experiment.

There’s a PurestDualPan already. It’s out, it’s free, it’s yours. There is also a BitShiftPan, which is a lot like BitShiftGain except you can control the sides independently. The first, PurestDualPan, is simply a dual pan in long double precision and dithered to the audio buss. This is unreasonable overkill but I’ve always been fine with that. I’m trying to get a sound and only have bits to work with. So in the ‘bit’ versions, they manage to get the sound of ‘utterly pristine and undiminished’ through gain changes by applying simple gain changes but only in 6dB increments.

In floating point, whether it’s 32 bit or 64 bit double precision, if you do that something special happens. A floating point number has two parts: a mantissa and an exponent. The mantissa is literally a tiny chunk of fixed point audio data (well, unless it’s just math data, but the same thing applies). The exponent? That’s how loud to play the first chunk. All the ‘bit’ plugins do is alter only the exponent, and pass through the mantissa untouched. Perfect audio… in 6.08dB increments.

For BitDualPan, that’s what you get… almost.

If you leave the sides hard-panned, the mantissas are untouched. If you swap the sides, same deal! If you set one side, or the other, to zero (silence) and then do anything at all with the other side, same deal. And then, if you pan in the sides (in 6dB increments, still) and leave both audible, that means both sides will be altered. They’re added together, so they’re a floating point add from two sources both of which were bit-shift gain staged, meaning the mantissa will be altered, and that means… what?

I honestly don’t know. I’m sure I don’t have the nicest playback system of anybody who uses my plugins, or the best ear: I’m just willing to try stuff and this becomes a new form of minimal processing. Instead of dithered multiplies, it’s summing a couple of floating-point inputs in such a way that there is only an add, nothing else. It’s certainly not gonna be any worse than any ordinary DAW summing, but does it gain anything from setting up the source audio so both sources are the exact mantissa of the input channels? It reduces the whole thing to a single add (per channel) and no multiplies, which are normally what you see in this sort of thing.

If you feel this one, use it in good spirits. It manages to cut the digital math to an absolute minimum but then apply it in a way I don’t often see, and it sits in the ‘dual pan’ role that I’ve long overlooked. And it uses the mechanic the Bit plugins enjoy: when you’re not touching the mantissa, you also don’t dither because the output’s already perfect. So, BitDualPan also doesn’t floating-point dither, to see if there’s a sound that comes out of the only mechanic being a floating-point add.

I’ll get right back to plugins that have more going on. There’s only so much you can do with ‘what is the sound of one floating point add per channel?’ But here you are, to compare with PurestDualPan. Now you can pick whichever you like best :)

Airwindows Consolidated Download
Most recent VCV Rack Module
download 64 Bit Windows VSTs.zip
download Signed M1/Intel Mac AUs.dmg
download Signed M1/Intel Mac VSTs.dmg
download LinuxVSTs.zip for x86
download Pi4VSTs.zip for 32-bit Pi
download Pi5VSTs.zip for 64-bit Pi
download Retro 32 Bit Windows VSTs.zip
download Retro PPC/32/64 Mac AUs.zip
download Retro PPC/32/64 Mac VSTs.zip
Mediafire Backup of all downloads
All this is free and open source under the MIT license, brought to you by my Patreon.

Date May 31, 2026
Author Chris
Comments No Comments

SoftClock3

TL;DW: SoftClock3 is a groove-oriented time reference.

SoftClock3.zip (535k) standalone(AU, VST2)
SoftClock3 in Airwindows Consolidated under ‘Utility’ (CLAP, AU, VST3, LV2)

You might or might not have heard of the first SoftClock. It came out shortly after I’d developed it because people desperately wanted to play with it, and before I was able to use it much. (still can’t, too busy with plugins.) It’s the metronome made of waving, wobbling tones where the beat is always a swoop of a (mostly) sine wave that goes up to a high point. Inspired by a joke metronome of dog woofs that proved strangely inspiring to play along to, SoftClock is the machine tempo with perfect regularity but no actual beat, where you have to place the ‘click’ of your notes wherever it SEEMS to be.

Version 3 is largely about taking SoftClock in a strange direction where it builds in Airwindows theories about tempo from 2020, and designs its wobbly guide sound around them, producing distinct wobbles based on what sort of tempo node you’re using.

But what on earth is a tempo node? Well, in practical terms it’s the thing that took away the whole ‘big beat’ and ‘snare pocket’ and ‘speed’ controls, replacing them all with ONE control that is simply ‘Flavor’. Set ‘flavor’ to zero, and you get a simple repeating pitch-waver to guide your tempo. It is not a ‘click’ but you’ll find it easy to sync up to it because the pitch-waver sounds like moving your body feels, and you can set a tempo by its regularity. The beat is where you feel the beat is. The waver is perfectly regular, so it doesn’t drift.

Set ‘flavor’ to 0.5, or even more, and things start getting a lot more interesting.

That’s because this ‘node’ concept of tempo covers a lot of bases. You can have ‘relaxed’ nodes where the groove mustn’t be too rigid or strenuous. You can have ‘tense’ nodes where it’s positively brittle, loaded with energy. There’s nodes that amp up the relaxed energy until it’s an infectuous groove… or nodes that take the relaxed energy and hang back, sneaking in attitude until the groove swaggers. It’s a continuum of groove, cycling between serene ease and jittery edginess, and it wraps around and around rising tempo like a barber-pole from the slowest to the fastest tempos. One of the Airwindows fans, Bo Danerius, worked out the algorithm and shared it, after I made a post outlining the theory: math webpages can do things like that given scatter-plots of data, and I’d sorted a whole bunch of hit records into these categories.

And now, SoftClock3 does it automatically: just set the tempo and it AUTOMATICALLY applies the speed, swing and wobble settings to deliver the correct pitch-wobble for whatever that tempo is. No guesswork. It even helps you tune in exact tempos for a song or riff: if you’re near the ideal setting, it’s likely that one direction will be ramping up speed and tension, and the other direction will slow the wavering and calm the energy. Either of these directions could be either faster tempo or slower: depends what node you’re already at. It ought to be a very natural way to find the correct tempo for whatever you like, by roughing in the general region and then exploring until SoftClock moves the way you need the song to move. Use the Flavor control to govern the amount of character that should be applied to the tempo: full relaxed or full tense don’t have swing, but groove and swagger do, and they’re adjusted by the amount of Flavor.

If all this sounds like complete madness, that is perfectly fine. It’s experimentation. We can’t all go ‘tell me what plugin you want and I’ll have AI make you one’. Some of us gotta deal with getting these big ideas, and pursuing where they lead, sometimes for years. It’s been six years since I posted the first thoughts on this stuff. SoftClock already existed. Why not run with the combination of all these things, and see what it does?

SoftClock3 is a groove-oriented time reference. It’s pursuing various theories nobody else has, in hopes of making a kind of ‘click’ that so perfectly represents entrainment (the way musicians feel time) that it can not only lock people in effortlessly, but also lead them towards the perfect tempo, through exaggerating the flavor of that tempo until it’s impossible to miss.

If it works, we can all discover a new thing together… and lay down some momentous grooves of all flavors and varieties. :)

Airwindows Consolidated Download
Most recent VCV Rack Module
download 64 Bit Windows VSTs.zip
download Signed M1/Intel Mac AUs.dmg
download Signed M1/Intel Mac VSTs.dmg
download LinuxVSTs.zip for x86
download Pi4VSTs.zip for 32-bit Pi
download Pi5VSTs.zip for 64-bit Pi
download Retro 32 Bit Windows VSTs.zip
download Retro PPC/32/64 Mac AUs.zip
download Retro PPC/32/64 Mac VSTs.zip
Mediafire Backup of all downloads
All this is free and open source under the MIT license, brought to you by my Patreon.

Date May 24, 2026
Author Chris
Comments No Comments

kCyberCity

TL;DW: kCyberCity brings live atmosphere to deep reverb.

kCyberCity.zip (644k) standalone(AU, VST2)
kCyberCity in Airwindows Consolidated under ‘Reverb’ (CLAP, AU, VST3, LV2)

If you count the number of Airwindows reverbs, even just the ones with ‘k’ in front, it’s pretty daunting. This journey has taken years. Most recently, I’ve been working on the basic technology of how the reverbs are even made (running a 6×6 Householder matrix fed by a 3×3 early reflections matrix) and refining the program that I run, night after night, to explore the space of possible delay combinations using genetic algorithms.

And still, the best discoveries are the accidents.

kCyberCity sounded, from the first, like rainy neon dystopian futurism. Which is to say, it screamed ‘Blade Runner’, but it doesn’t actually have a Lexicon algorithm in it… it’s way more complex than that. It sounded like what I imagine Blade Runner sounded like. Specifically, I got a sense not of a big room, but of rainy city streets, washes of echo return, a particular sheen that felt like neon reflections on pavement sounds.

But what really helped (in more practical terms) was developing a ‘room tone’ method that involved not only feeding ‘VoiceOfTheStarship’ noise into only the regeneration part of the reverb, but also coming up with a way to modulate that, where before I’d tried things like ‘Discontinuity’ in that position. What I got was special: where I got a hot intense sound in a reverb like kGuitarHall2, this time I got the ability to disrupt the deep reverb field as if listening into a live environment, with wind and atmosphere. And two separate ways to manipulate that, for defining the apparent size of the reverb space and for defining how still (or un-still) the air was in there. Oh, and also a feedback nonlinearity that tweaks the whole character of the space.

This is the first truly live Airwindows reverb, but won’t be the last. Turns out that even if you drop the ‘room tone’ right down to what is realistic for an acoustic space, it still makes a huge difference in believability. A lot like dither, really, but considerably more obvious. You’ll have to use BitShiftGain (or compress really hard?) to get it to be obvious, but you don’t have to. The room tone is heightened when regen is turned way up, if that’s your intention. In normal use, think of it like a sort of dither, but for spaciousness.

There’s quite a few more along these lines, especially if you count the smaller spaces I keep getting asked for. Those should respond well to this new tech and convey a powerful sense of what they are. It’s just that starting with a spectacular sci-fi fantasy is too much fun to miss :)

Oh, also, you can adjust Position on the fly now, or automate it. Also also, the Pi5 build now specifically supports 64-bit Pi, so the Pi5 build of kCyberCity will run on 64-bit Pi, plus also every single other plugin in that collection (close to 500: ALL of them) also now runs on 64-bit Pi. The Pi4 build is still built on a 32-bit Pi 400 to support those humbler tinyboxes, which should also run this just fine :)

Airwindows Consolidated Download
Most recent VCV Rack Module
download 64 Bit Windows VSTs.zip
download Signed M1/Intel Mac AUs.dmg
download Signed M1/Intel Mac VSTs.dmg
download LinuxVSTs.zip for x86
download Pi4VSTs.zip for 32-bit Pi
download Pi5VSTs.zip for 64-bit Pi
download Retro 32 Bit Windows VSTs.zip
download Retro PPC/32/64 Mac AUs.zip
download Retro PPC/32/64 Mac VSTs.zip
Mediafire Backup of all downloads
All this is free and open source under the MIT license, brought to you by my Patreon.

Date May 17, 2026
Author Chris
Comments 3 Comments

BezEQ3

TL;DW: BezEQ3 refines the concept into four vivid, unified bands.

BezEQ3.zip (516k) standalone(AU, VST2)
BezEQ3 in Airwindows Consolidated under ‘Filter’ (CLAP, AU, VST3, LV2)

Persistence has its rewards…

The idea was to see if BezEQ could be expanded into something I can use in a Console, for which I like having four bands and frequency sweep controls for each. But the delay lines built into BezEQ2 had to go, because I’d also want zero latency… and that would produce a phasier response, perhaps interactions between bands that would make it behave less normally. But normal’s easy to come by, SmoothEQ2 exists, so what if we could get something worth having that’s less normal?

Enter BezEQ3. There’s more improvements on getting it to filter smoothly, there’s that needed 4th band, and the sweepable middle positions to the bands, but it’s still very much BezEQ. It’s refined that unusual tonality down, and is still using a filter topology where, set flat, it’s literally perfectly transparent.

And then it’s perfectly transparent for any other setting, no matter how extreme… but how you GET to those settings is something else.

Because removing the internal delay lines and making it be zero latency, derailed the filter bands in some peculiar way. If you boost a frequency, you’re cutting something else. Everything’s kind of interactive. It’s not going to behave nicely if you think you can treat it like shelving filters with neat markings. No point EVEN marking it. This is not SmoothEQ2, which runs on DF1 biquad filters like a normal plugin.

Instead, you voice it by using the bands and sweeps as suggestions, and you listen to what it’s doing. Nothing will get a frequency band to separate out and act like a ‘frequency band’, and curiously nothing will get frequency bands to come forward or fall back (like FatEQ) either. It acts, more than any other EQ I’ve ever heard or made, like no EQ had been done, and like the sound was that way on recording.

It’s just that you can revoice things to a ridiculous degree and still have it sound like one coherent piece.

This plugin will be very polarizing. You will likely love it or absolutely hate it. I suspect some of the other old-schoolers will like what they hear. I’m almost sure anybody with extensive modern-production expertise will be frustrated and confused. So very many people think in terms of taking all the sounds and decomposing them into distinct pieces and fitting them together like some mixing jigsaw puzzle. What are you going to do with an EQ that refuses to separate out anything, where you can never turn a sound into this band and that band and holes where the bad audio used to be?

Meanwhile, the old school practitioners of mixing black arts are licking their chops and can’t wait to turn a much smaller number of tracks into something huge and compelling… sneakily heightening rawness and energy, without ever seeming like they are EQing. It’s more like voicing. Nothing needs to be aggressive, just voice everything until it hits right and sits in the right places, and the top-to-bottom, left-to-right, mixing jigsaw puzzle can stay in the box. With suitable use of compression and saturation you could use this to mix a whole thing in mono, sounding unequalized, and it’d still hit like a ton of bricks.

BezEQ3 is currently on track to be the primary EQ in ConsoleX3. Many people will just get mad at it, most likely. I’m just gonna have to show people what this is about… but if something happens when you try this plugin, and it gets exactly the sound you crave and you want to use it on everything… well, you know who you are, and you’re welcome :)

A follow-up note: I’ll be working on getting a 64-bit build for the Pi 5, might be leaving the Pi 4 build as 32 bit for the use of even older Pis.

Airwindows Consolidated Download
Most recent VCV Rack Module
download 64 Bit Windows VSTs.zip
download Signed M1/Intel Mac AUs.dmg
download Signed M1/Intel Mac VSTs.dmg
download LinuxVSTs.zip for x86
download Pi4VSTs.zip for the Pi 4
download Pi5VSTs.zip for the Pi 5
download Retro 32 Bit Windows VSTs.zip
download Retro PPC/32/64 Mac AUs.zip
download Retro PPC/32/64 Mac VSTs.zip
Mediafire Backup of all downloads
All this is free and open source under the MIT license, brought to you by my Patreon.

Date May 10, 2026
Author Chris
Comments 4 Comments

Chris

BitDualPan

SoftClock3

kCyberCity

BezEQ3

human-made bespoke digital audio

Kinds Of Things

The Last Year

Patreon Promo Club