Menu Sidebar


Airwindows is one guy: me! I am an audio hacker and computer programmer from way back. I seek only to continue my life up here in Vermont, inventing things and putting them onto the internet, sometimes for free and sometimes for pay. My hope is that people richer than me (i.e. most people) don't rob me. My other hope is for another cup of Aeropress. One out of two ain't bad!

Guitar Conditioner


Replace your Tube Screamer (plugin OR REAL DEAL) with this plugin, to get ‘scrunch’ and subs conditioning, but without the sponginess that even hardware TS9s have.

Guitar Conditioner

I designed this plugin when I saw what Slew was capable of. It uses a combination of Slew and Highpass in a sort of parallel matrix arrangement, with heavy use of my Highpass’s unusual tone shaping features (extreme lows get Tight tone shaping behavior, and a boosted mids circuit is set to Loose rolloff for the proper texture). Both aspects hit their own Slew at different levels and with different voicings, causing a distinct tone quality.

This is not a Tube Screamer plugin at all. It’s an alternate tone for doing similar things but with more sonority, focus and impact. It’s intended for Djent style sounds and people who can play much, much better and more aggressively than me :)

It is free. Get more like this by supporting my Patreon, because lots of other people can give you ‘imitation Tube Screamer’ but Airwindows is for trying to give you ‘beyond Tube Screamer’ through understanding what, in the conditioned sound, should be distilled and extracted. Happy Djenting :)


Channel4 is the latest Airwindows port. This plugin has always been popular: once I had someone shoot it out in a blind test against commercial rivals, and it trounced some heavily hyped and widely known plugins!

At the time, I was amused and distressed, because I would have liked to see one of my $50 plugins in that comparison. To use the freebie seemed like missing out on some helpful sales.

Now that I’ve switched over to Patreon for Airwindows’ only means of support, it doesn’t matter and all the plugins are steadily becoming free :)

Channel4 uses a very good-sounding saturation function, a calibrated slew clipper and a highpass to approximate the tone coloration of a recording console. (the plugin Console4 approximates the space and depth on the mix buss, which is different.) Because the bandwidth-limiters are set up based on measurements of real gear (through impulse responses), Channel4’s console type is labeled ‘Neve/API/SSL’. It doesn’t actually contain any of those things or any convolution kernels, but the presentation of the sound ought to suggest those voicings.

There’s also an argument for doing it that way, because in many cases you can’t hear those limitations unless they’re being way overloaded. Channel’s a very clean, pure ‘coloration’ plugin, especially with the drive kept at 50% or under. It’ll voice extreme sounds the way it’s labeled, but you would struggle to consistently set plugins like Slew to those targets because mostly you can’t hear what it’s doing, especially on normal/musical sounds.

I’ve got a big list of further free VST ports, and am looking into some bugfixes for PC hosts that aren’t handling generic VST2s right, and I’m actively supporting users on several forums and email and Patreon messages, so if you’d like to see this continue to expand, please support the Airwindows Patreon, that is now the way things are done around here :)


My first Windows VST (and Mac VST) port was Console4, the very latest version of the best selling plugin I’ve got (now free, supported by Patreon.)

The second one is the very first plugin I made for Airwindows (or at least the first one listed on this website).

Slew is a simple plugin to let you do slew clipping. You can use it to morph and transform percussive sounds or hats, you can give a unique and interesting ‘grind’ to instruments or glue stuff together into a retro, old-school-sampler, grungey grind, or you can use it on things like reverb sends to really amp up the sense of distant loud sound in a room or space.

It’s free, and in the zip file are an Audio Unit (.component), a Mac .vst, and 32 and 64 bit Windows VST .dlls for you.

Hope you like it! Lots more is on the way, and that’s not even counting when I reach the Patreon level that I start putting out the for-pay plugins. I have lots of Audio Unit freebies for people to enjoy, and I’m busily porting them. All of them ;)

Here is an extra video about using Slew on ITB heavy guitars, to get Tube Screamer-like qualities but in a new way.


Console 4 is the most recent version of basically my flagship Airwindows plugin. It works through a channel plugin on every sound source, feeding at unity gain into the buss plugin which overrides the digital summing with Airwindows summing (more internal space, more depth, and with Console 4 it now has glue making the top-end more listenable).

You can also watch the half-hour long introduction video that goes into a lot more detail about how it’s used, how it works, how you make ‘big kicks and snares’—Console doesn’t let you crank any given channel up to full volume, but it’s designed to layer stuff so if you need giant sounds the best way is to make them composite sounds, everything layered from separate channels in the mix. Rather than layer samples and put them on a single loud track, keep the layers accessible in mix. That does mean if you wanted things like sines to be superloud you’d have to get creative, but the principle’s clear.

Speaking of principles…

You may notice no demo link, and also there’s no Kagi shopping cart link. There’s a reason for that. Kagi went bust Sunday. I may still get my last two months of sales out of whoever’s divvying up the assets, but as of this Monday you cannot pay me for my work through Kagi, because there is no Kagi. They served me for close to ten years, always with perfect efficiency and scrupulous honesty, but the commission off my work wasn’t enough to keep ’em going.

I could go find the newest-trendiest shopping cart e-commerce people, but when I started shopping-cart shopping I found them all horrible, tacky hypemongers offering to do things like find people who’d left a cart un-checked-out, and spamming them with reminder emails like ‘Hi, I’m your shopping cart and I’m looonely! Do you miss me? I miss you! Surely you just forgot me?”

I ran away before I threw up all over their e-commerce portals. After all, I have never spammed or bugged people, never advertised, and what’s more I give people free updates for as long as I live, none of which ‘makes sense’ in this happy future of badgering people for every cent, preferably by hyping them into a rental arrangement and DRMing the stuff until it’s nearly ready to explode all by itself, never mind when the rent is due or the authorization servers are having a bad day. (but I digress)

There’s something completely different I can do—something I would never have done, except Kagi went out of business and I have no reason not to be completely rebellious and flip the marketplace table.

Ever heard of a thing called Patreon? It’s not for discovering new artists. Really, it’s more of a… payment processor. For people who are already well established, who are appreciated for what they do, and who are busy devoting their lives to giving the world something for ‘free’ (like comics, or perhaps music, or art). Much like I already do for existing customers: I’ve promised all Console owners that they will have all Console updates for free.

There you have it. I am going all in on Patreon, and that is the full release version of Console 4, with the new Mac and PC VST builds, for free. Please remember this when I have figured out the details and started my Patreon. From now on, I will be relying completely on that to survive. I’ve seen more than one person (for instance, Vechs, or Jim Sterling) who are doing great with Patreon, usually because they too are rebelling against some commercial thing and making a bid for total freedom from obligation.

From now on all Airwindows plugins will be ported to Windows and Mac VST as well as being Audio Unit, and they will all be free from now on, and if I get enough Patreon support I will release all source code under the MIT license and document it as I go so that everybody can use the tools and concepts I’ve built to create their own software. That will be my legacy, and if that ain’t a worthy Patreon goal I don’t know what is.

This includes old versions of plugins for when people preferred a certain version, so the total number of plugins to cover is over 250. That means if I do one a month it will take me more than 20 years. If I work like a maniac (well, more like a maniac) and put out one a week, that’s still around five years just to turn Airwindows into a sort of audio plugin library and DSP school. And it will all be AU/VST with VST covering Mac VST2 and Win VST2 (built on older OS versions so the plugins work on the very widest range of hosts). I feel this isn’t a bad strategy because if I set up the Patreon so I can actually survive on it, I can make the rate of these ports (and free releases of the existing AUs) conditional on whether I was able to eat that month, which seems fair. :)

That’s not counting new research and new plugins… but I have a lot of VST plugins to do, to catch up. Over 250 of them. I’d also ask pirates and haxxors (who don’t have any VST airwindows but what I make, and don’t have most of the AUs) to please leave this ‘making them all free’ process to me through Patreon and my own website. If all goes well, all the plugins will be free in the end (even with source code!) so if you could not mess this up I’d be grateful. ;) leave it to me, please.

Oh, and the VST versions do ‘double replacing’, which means they noise shape to the 64 bit floating point buss. Technically that makes them higher sound quality than Apple’s CoreAudio can offer, though I promise you really won’t notice (and the VST versions also do 32 bit ‘replacing’ so they’ll work on all hosts)

Consider Console 4 an advance on this new concept. If it works, and if I can live for five or twenty years doing it, every possible Airwindows plugin will be part of everyone’s toolbox and the code will be out there making people’s products better.

Private to Native Instruments: hey, maybe you guys might want to chip in just to meet whatever the threshold is for ‘release free ADClip with source code under MIT license’! I will be letting people pay to mess with the release schedule directly, so past a certain threshold people will be able to single out specific plugins and fast-track them and get the source opened. We didn’t come to an agreement for a one-time no-royalty payment, but now might be your chance! ;)

Oh, also: Patreon :D (I updated the link once the site was ready, and then I started making all the words links, and it got funnier and funnier. patreon patreon patreon! OK, I’m done ;)


ADClip6Demo is a universal binary Audio Unit plugin for loudness maximization. While it can be used for clean peak-clipping, and is great for that, if the words ‘competitive loudness’ make you sad, stop reading now.

Still here? OK. Here’s the deal: ADClip was already one of the most formidable weapons for loudness. The last version, ADClip5, introduced some new ways to monitor what you were doing, and a pesky high frequency bug that bit a few people who were using it on VERY bright content. That bug’s squished, which is one reason the new version’s come out.

Another reason is this: there’s a new monitoring mode that compensates for input gain. So if you’re boosting inside ADClip, you can use that to listen at equal loudness to the results and be sure you’re not getting fooled by simple loudness increase (you should be able to hear where the sound gets smaller as it’s squashed by clipping), plus using that it’s easy to bypass the effect and compare it directly with unclipped, unrestricted output.

But the real reason is this: the sound doesn’t get smaller when you crush it. It gets bigger. ADClip6 under heavy clipping has at least a full octave of deep bass under what ADClip5 could do, which was already more than most loudenators can offer. You can plainly hear this in the demo, and you can dial it in any way you like: exaggerating it, or pulling back the Subs Retain until you have the full scale of the music intact, but at a previously unreachable loudness.

If that wasn’t enough already, the main signal path plus the Subs Retain path also use a feature called Fatten Body, to transition more gracefully between clean audio and the smashy smashy. You can use it to finetune the fullness of the resulting output. To get the cleanest possible output in ‘safety clipping’ situations set it to zero. If you’re mastering for bigness, use it to balance the body and fullness of the track with the loudness and punchiness gained through driving ADClip6 hard.

And there’s a Soften Clips control, which can adjust how much ADClip6 de-glares the clipped highs. All of these work fine at their default settings of 0.5 but you should tune them to the needs of the audio you’re processing! Be careful as ADClip6 doesn’t react like normal clippers, it keeps going and going. Use subwoofers to monitor what it’s doing on the low-end, and compare the volume compensated version with the bypassed, unclipped version. Remember, added punch and openness is good, but ‘twice as loud as anybody else’ is just annoying even if it sounds awesome to you! Please don’t use this tool for evil. <3

ADClip6 declares no latency so you can track with it as part of a zero-latency DAW or live setup, but it will delay the signal by exactly one sample: I figure that’s not going to affect performance timing, and on the 2-buss it won’t affect relative positionings of tracks. We might be able to see better DAW performance when all the plugins used are declaring zero latency, so I’ve made this change. For use in submixes and places where you’re smashing tracks with a clipper to get specific tones, I recommend looking at the Airwindows plugin OneCornerClip, which is more of a waveshaper and delays zero samples. ADClip needs to also soften the exits from clips, so it can’t do that. Or, you could just use it and deal with the one sample internal delay (for instance, on an overall drum buss).

ADClip5 is $50.


CompresaturatorDemo is a universal binary Audio Unit plugin (PPC/32/64 intel) that uses a completely original algorithm: not really a distortion, not really a compressor, it’s kind of both! Dial in qualities unachievable with normal dynamics processors!

Here’s how it works. It starts with the classic Airwindows ‘softest possible saturation’, as found in PurestDrive. This is a distortion that clips completely at 0 db, is still having a tiny effect even as quiet as it’ll go, and has a perfectly smooth transition between these states (a sine function).

We take what’s distorted off the wave (always present, but more as the gain increases) and we run it into a time buffer, like a storage tank. Rather than running a time constant in the usual sense, the turn-down factor is just the average of all this. No time constant abruptly turning down the signal as you exceed a certain threshold! There is no threshold, it’s perfectly fluid and smooth. Not only that, the range (given in the ‘expansiveness’ control) chases up and down depending on how intensely driven the signal is.

What does this mean? Under normal circumstances, compressor gain gets modulated by the signal in an obvious way, and it changes the tone of the audio. With Compresaturator, as the time range is getting shorter (one sample at a time) it’s making each sample count for more. But then, on the release as the time range relaxes, the change in gain level is only whatever the new sample is (could be silence!) and on the trailing edge, the time range is expanding, so the gain’s not being modulated by trailing edge samples either.

What this means is, Compresaturator’s gain stage is smoother than anything. It just doesn’t modulate to audio waveforms at all, especially at higher ‘expansiveness’ values. The gain manipulation is totally fluid, and sounds that way. (It’s also very, very CPU-efficient)

Here’s the trick. This can only work in a saturation plugin! Without that, you’d have crazy transient spikes as the gain stage refuses to kick in with the speed of a normal compressor. Not only that, by design this can only clamp down on the tail end of a wave, so it reshapes everything to hit with more dynamic impact and thickness on the front, and then cleans up right away. And since it’s the softest possible saturation, even the ‘cleaning up’ is extremely fat and thick, extra up front and punchy.

So, this is an extremely simple and well-designed process that only works because all of its elements work together. It’s perhaps the ultimate ‘glue’ processor, conditioning the tone even at zero boost, or if you attenuate going into it. Compresaturator makes stuff impactful, dimensional, makes tracks jump out of the speakers and sound more alive, and it’s truly ‘the bomb’ but it’s all those things at a level of purity and sonic honesty that can stand up to complete overkill and smashing stuff aggressively into the plugin. Compare beating up on Compresaturator, and how it sounds when you make its effect exaggerated, with other current plugins which do “impactful, dimensional, wide, and exciting”. Hear the sonic signature of what the plugins do when you hammer them. Then dial it way back (as one’s frequently advised to do) and consider how much better the ‘glue’ effect is when you know it’s not a bad tone coloring. That’s Compresaturator: the latest in tone hyping and impact enhancing, making the front of the wave hit with real solidity, but keeping your textures and tones intact.

Compresaturator is $50.


StarChildDemo is a universal binary Audio Unit plugin (PPC, 32 bit Intel and 64 bit Intel) that does a variation on the old Ursa Minor Space Station, but different. It’s a high purity stereo-only rendition of a raw and grainy antique digital spacemaker, all vibe!

Here’s how it works. Back in the really crude old days there were only delay lines. StarChild is doing a series of delay lines in a primitive, fixed point digital delay with a subtle pitch shift applied in an interesting, crude, grainy way characteristic of these really old processors. Is it low bit, for that low-res grit and crunch? Yes and no… that’s what the ‘grain coarse/fine’ control is for. It uses sustain duration to give roughly the amount of ‘hang time’ on the effect, but then the grain control determines how densely packed the delay taps are.

It’s implemented in such a way that coarse grain is insanely CPU efficient, and the fine grain is up to 162 distinct delay taps (more than antique digital processors could have done directly) and still very efficient for modern DAWs. The spacing of these taps is done using a set of hard-coded prime number delay times to keep overtones from reinforcing, so the only coloration you get is the overall ‘reverb’, and it’s done at four points in stereo to give you a spacious effect.

What does it sound like?

Like nothing else in this world. There’s a really strong ‘droning’ quality like certain Eno multitap delay effects, but concentrated on a fundamental ‘ring’ without affecting all possible harmonics of the sound. The pitch shift adds another level of strangeness and fuzz, barely noticeable but you’d miss it if it wasn’t built into the algorithm. There’s an airy quality that can also be a grungy quality at coarse grain… and the onset is softer than the rest of the sustain, making it a weird and amazingly unnatural sound that still blends with the dry signal in a very special way. And the way the attack is designed, the effect always ‘blooms out’ of a distinctly low-bit onset, which blurs and fuzzes out the attack of the ‘reverb’ keeping it from interfering with the dry signal. It’s like there’s always a little noise built into the attack of the thing, at whatever level you use, and the grain control determines how much of that you’re hearing.

It applies a super-strong flavor, but in a strangely useful way that doesn’t ruin the tone, just makes it richer and weirder. Both of the controls, sustain and grain, work to adjust the rough pitch center where all the jittering delay taps sit (they shift in ripples, one sample at a time for maximum scrunch, and one at a time rather than all at once for added texture).

For that hybrid of old-school and ultramodern electro madness, StarChild can’t be beat. It’s a true secret weapon, a bizarre little algorithm that sits sneakily in a mix just adding its alien mojo. You can have it barely audible and still it has a huge effect, or you can crank it right up and whack people with StarChild’s peculiar skronk, a texture that’s sure to jump out of any mix. You can make robot voices, electro drums, techno soundscapes, all with a glorious fuzzy digital obnoxiousness that screams ‘oldest of old school hardware box’.

StarChild is $50.


AuraDemo is a universal binary Audio Unit plugin (PPC/32/64 bit AU, for Mac) for dialing in the sparkle or sonority of a sound. It gives a kind of presence peak never done before in software, along with attenuation past that peak. Defaults to natural tone, or exaggerate it for hype!

Here’s how it works, and why you can’t do this by using the same response curve (seen in the video) on any EQ, digital or analog, no matter whose it is.

Aura’s not a complicated peak-and-notch maker, doesn’t operate with normal EQ algorithms or even ‘analog modeled’ EQ algorithms. It’s almost a Purest-line plugin. The ‘circuit’ being done in code is very simple and involves very few operations, much like a high-end preamplifier that works using very few discrete components. The operations it does do, however, increase the richness and sonority of ALL the sound, not just the stuff around the ‘aura’ voicing.

It averages out the rate of change OF the rate of change OF the waveform, and then lets you blend that result with the dry signal. That’s it, but nobody else has this (it would be extremely obvious if they did). One possible reason is the algorithm’s tricky? It took some real hacking to get it to act right as you adjust the voicing control: it doesn’t automatically work like a DSP cookbook EQ algorithm. I didn’t even know what it would do until I got the thing working and started to experiment.

Why’s it called Aura? Because using it, you get to wipe out information above a certain highlighted frequency (it’s sample-rate dependent so the voicing will tend to emphasize the highest reaches of whatever sample rate you’re using) but KEEP the information if it’s legitimately part of a waveform. Aura only keeps harmonics, and throws out noise and garbage. And it’s a very finely tuned cutoff, a real scalpel for singling out specific overtones. It does in a very simple algorithm what would take dozens of poles of DSP filter to not even do, and makes it sound amazing.

Consider that for a moment. How could you tell a steep lowpass, ‘get rid of only the stuff that’s not part of the harmonics of my waveform, okay? If it’s part of the big waveform, please keep it kthx, otherwise axe it. Oh, and give me a really steep notch over that because I want to take a nasty overtone out, but I don’t like the sound of really high Q digital EQ so leave that plasticky tonality out please. Oh and one more thing. Please run zero latency so I can track into you’.

You can’t, regular EQ (especially digital) tends to show the strain when you make it do crazy things like that. Regular digital EQ is better off doing broad cuts very cleanly. But Aura isn’t even an EQ algorithm, it’s something much simpler and this is what you get out of it: by its nature you get this tendency to hype and retain harmonics of musical waveforms, that’s at the heart of what it does. By its nature it does potentially extreme things without sounding overprocessed because it’s not overprocessed: it’s very simple in design. And it’s zero latency, just to drive home how different it is from typical plugins. It provides definition and intensity/sonority of tone beyond any EQ, for this one common EQ-like behavior.

The video shows it working on sounds from a Korg synth, such as strings and piano and banjo, but did you notice it’s also on my headset mic? To evaluate what Aura means to you, I’d advise grabbing the (Audio Unit) demo and trying it on sounds you normally use. It ought to be killer on any sort of vocals, particularly if there’s a high presence peak that’ll work for the sound. You can increase the voicing and bring it down into that high-middy obnoxious range that wouldn’t work for airy sounds, if your sound is darker like a piano or perhaps a bass, and it’ll refine the mids and give you loads of definition and articulation. And of course, any guitar sound, particularly heavy guitar sound, can become amazing through using Aura to dial in just the right presence peak slightly higher than the guitar speaker naturally puts out, so it extends the range while also notching out artifacts above that peak. You can get a whole range of flavors on cymbals, you can invent new sonorities for entirely digital synths… treat it as a kind of lowpassing, a very useful and common class of effect, but one that brings out your harmonics rather than dulling them and choking them off.

If I can make the same thing happen in a highpass, that will become a free update for anyone who got Aura (because it would be the most natural and best thing ever to bring Aura the ability to ‘trap in’ sounds on both frequency extremes). For now, I bring you the ultimate ‘lowpass kind of thing’ that’s technically not an EQ, but behaves a lot like one.

Aura is $50.


OneCornerClipDemo is a universal binary Audio Unit plugin (PPC/32/64 bit AU, for Mac) and one of the coolest things I’ve done in years.

It’s basically a special clipper where you can RADICALLY shape the tone of extremely distorted output without so much as touching any samples that aren’t clipped. The voicing control has a huge impact on the tone of the thing. The secret is this: once you clip, the waveform starts from where the last good sample was and begins rounding off the corner until the signal passes it going the other way. This means high frequency stuff gets well damped down, and it means you can reinforce low bass under conditions of super-obscene hyperclipping.

Then, exiting the clip, you get one perfectly sharp corner (much like if you were lowpass filtering a square wave). Hence, OneCornerClip. The corner becomes part of the waveshaping and stops sounding as much like digital clipping, because there’s only one per wave rather than two, and it’s in a more regular spot rather than being pairs of corners spaced at arbitrary intervals.

That part is true even for massively oversampled clippers: they’re still producing corners that are irregularly spaced depending on how far into clipping you went. OneCornerClip doesn’t do that, so it sounds like LOUD! not like plugin clipping. It feels like you’re cranking some physical device up, but without altering any sample that’s not clipped, without any kind of convolution or EQing or blurring. Clean samples are bypass, period.

For anybody who is trying to get fat EDM grooves to smash through all imaginable loudnesses forever, this is the tool. Dial in the voicing real low to bring out the kicks, you literally can’t lose kicks through smashing this thing. It depends on how you set it.

For anybody who likes taking a mix element like drums and going all NIN on them, producing an intense driving beat, this is also for you. The default voicing of 0.618 is quite good here but you can push it higher for a real aggressive rowdy top-end, or lower to produce drum-machine like effects on cymbals. As the voicing gets lower, cymbal overdrive will get more ragged and synthetic.

At very low or zero voicing settings, you get a dirty, grunged sound because you’re getting digital flat-topping but at many different output levels, never the same one. I’m pretty sure nobody else can do this yet and it doesn’t exist in anybody’s synths either. At a voicing of 1.0 you get literally raw digital clipping, nothing more: you can always hear what that’d be like, just by moving the slider.

Lastly, OneCornerClip can be used anywhere anytime: it’s very efficient, is coded ‘N to N’ so it can work on anything from mono to stereo to 7.1 surround channels (most of my plugins can do that!) and it has zero latency and zero delay. It’s like an analog device that does all this on the fly without getting bogged down in calculating and processing. Like the Purest line of plugins, it runs an 80-bit buss when it’s doing its thing, and when it’s not clipping it passes through the input value totally unprocessed, like bypassing the plugin completely. Heavily oversampled clippers can’t do that, and can’t match the rawness and pureness of OneCornerClip passing through normal audio.

I’m real excited about this one. I’m going to have a lot of fun with it, I think, and so will a lot of other people. Play with the demo and you’ll soon discover some amazing things are possible. And enjoy the video! I did! :)

OneCornerClip is $50.

Older Posts


handsewn bespoke digital audio

Kinds Of Things