Menu Sidebar
Menu

Chris

Hi! I've got a new plugin you can have! These plugins come in Mac AU, and Mac, Windows and Linux VST. They are state of the art sound, have no DRM, and have totally minimal generic interface so you focus on your sounds.

PurestWarm3

TL;DW: PurestWarm3 is a subtle tone shaper and warmth adder.

PurestWarm3.zip (503k) standalone(AU, VST2)
PurestWarm3 in Airwindows Consolidated under ‘Subtlety’ (CLAP, AU, VST3, LV2)

It’s been some years since PurestWarm saw any attention. It’s always been a bit tricky: proper second harmonic distortion is ‘warmth’ but it’s very hard to hear. But that’s just the start of it…

Every version of PurestWarm ever (OK, so that’s just the original and v2) has had a flaw that I didn’t fully understand in 2022, and have come to respect. Freya Holmer’s video on spline continuity tipped me off, but I learned about it years earlier with an ill-fated Channel ‘upgrade’ that I had to roll back as if it was ‘New Coke’. People respond to sudden changes in derivatives even when the waveform seems to have perfectly good continuity.

The crossover through zero in an audio waveform counts.

So, the previous versions of PurestWarm do an abrupt switch between ‘dry’ and ‘effect’ at the zero cross. That’s nearly as bad as what a sine does (abruptly changing from positive to negative curvature at zero) and even though it’s fairly hard to hear the largely even harmonics this produces, it’s still a limited approach. And so, PurestWarm3 is here… but what’s it doing that’s different?

An odd little use of ‘FMA’ (fused multiply-add) to produce a super high resolution version of a continuous asymmetrical distortion. One that can scale up and down to produce more or less of an effect. One that’s already been heard in RetroBass, in ChannelX… now you’ve got it standalone, in a stripped-down and minimal plugin that only does that one thing.

And it’s so much purer and smoother than even the previous two, that you damn near can’t hear it at all.

That’s the trouble with these. PurestWarm3 actually throws two harmonics, a second and a fourth. It doesn’t matter much, you still can barely hear it. I’ve set it up so it can be exaggerated wildly, just to be heard. The challenge there (not present with the previous two) is, this is a continuous algorithm. It just reshapes everything. So what it takes from one side of the wave, it adds as peakiness to the other side, and so it will distort enthusiastically if you push it too hard, and then if you don’t push it too hard you kinda can’t hear what it’s doing, much.

Add to that the same derivative BIP tech I’ve been developing, and it really really stays out of the way sonically. A puzzle.

So, this is not really a ‘slather on thicc warmth!’ effect. If you have sounds that are already asymmetrical, such as certain basses, it can reshape those to ironically make them LESS warm but cleaner. Or, you can exaggerate the effect and make it more striking. Push it too far and you’re messing with a kind of expansion that won’t play nicely with mixing.

I hope to make use of this in a future console plugin where it’ll convey the exaggerated transient spikes of certain late-70s noise reduction, so it’s not like there’s no use for it. Just don’t expect an ‘EQ-like’ activity out of PurestWarm. It is rearranging energy, almost losslessly, to be a subtly different shape, and leaving next to no trace it was ever there. To some of you, this might be of great interest if that got your attention…

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.

ChannelX

TL;DW: ChannelX translates Channel9 into a profusion of wild experiments.

ChannelX.zip (545k) standalone(AU, VST2)
ChannelX in Airwindows Consolidated under ‘Tone Color’ (CLAP, AU, VST3, LV2)

Everything’s the same, and nothing is the same.

It’s quite hard to tell what will happen with this plugin update. To surface appearances it’s exactly the same as Console9, right down to the slightly odd choices like a (lower slope) interleaved IIR filter for highpassing. The controls and what they do are exactly the same. It still has named ‘modes’ it operates in, still has a drive control that produces the Spiral saturator at 100% and Density at 200% and crossfades from dry to Spiral to Density, linearly. This is what Channel9 has had since 2021.

It’s not 2021 anymore… and Chris from Airwindows has been responding to the pressure of vibecode guys flooding the plugin market by trying new stuff more often… based on the existing Airwindows codebase, and pretty much nothing else. And Channel’s number came up. (technically, letter, since X can be used for 10)

Longhand was an experiment in using a Taylor series to make a sine function ‘by hand’ in case building it out of a single, continuous algorithm sounded better than the stock, optimized form. It seemed to sound nice. Both the sine part, and the Spiral part of ChannelX are now made that way, longhand, at very high precision.

RetroBass came out just last week! It involves modulating even harmonics into the sound. ChannelX now includes this in its highpasses, but still uses the dielectric modulation from Capacitor, and also still uses interleaved filters, which RetroBass does NOT use.

A stack of other plugins are being updated with an experiment I did inspired by elliptic BLEP antialiasing! But where that is established as a way to flawlessly synthesize a filtered result to any degree of high-order filter ripple, my experiments veered off to produce a tweak that operates over ONE sample, not to delete aliasing but to doctor it in such a way that high frequencies remain locked in place and do not ‘come forward’ in the way aliased highs seem to do. Totally other thing, seems to work, zero ripple or blur or SRC-like qualities. The directness is incredible. (people don’t seem to go for this quality very often in 2026, which might be as good a reason as any to pursue it)

So ChannelX is a wild mix of very new experiments under a familiar surface. There have been times I freaked out over how good it sounded compared to Channel9. With a bit of luck this bodes well for new plugins coming up. We shall see.

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.

RetroBass

TL;DW: RetroBass is a special kind of highpass.

RetroBass.zip (509k) standalone(AU, VST2)
RetroBass in Airwindows Consolidated under ‘Filter’ (CLAP, AU, VST3, LV2)

It started with a second-and-fourth-harmonic generator I stumbled into (never fear, you’ll get that too). Though it was a little more complex than the simplest second harmonic generators, it still had the unmistakable signs of them: you could mess with the settings until the output clipped like mad and still you’d not really hear what it was doing. The asymmetry has tone qualities, is warm and all, but at such low harmonics it’s just so hard to hear.

Causing me to wonder: can you use that to take away low frequencies like a filter?

So, combining it with a simpler version of the highpass filter I use in Channel, I start experimenting (on a livestream) by modulating it with the same IIR stuff I’m subtracting to make the highpass. Not only that, this asymmetrical even-harmonics thing is using the aliasing-manipulation hack I came up with for normal distortions just recently (and that is buck wild because I invented that on purely hard clipping, but it generalizes) and the next thing you know I’m figuring out whether it’s best to bring it in directly, or inverted.

And I observe that it produces this funny double bump of a distortion, but one way around the initial lobe of what was once a sine wave normalized to 1.0, is bigger, and what follows is both smaller and a tighter shape. It’s acting like it’s taking energy in the oscillating wave, and running out mid-oscillation. This is what analog circuits do, when their coupling capacitors run out of steam. It’s just being done in a peculiar way.

And dialing it in using the test waves, and then trying it out on lots of different things… I discover that it’s RetroBass. It’s better than anything else I’ve had, for specifically making stuff bass-starved in a retro-seeming way. It sounds like old gear not designed for modern subs.

Not only that, if I push it farther, it starts to get into those hot harmonic-rich zones that make some old vocals sound so intense. It’s like a treble zing that is done through weakening the fundamental. Very handy.

RetroBass will continue to be useful in things like retro consoles, but right now it’s a standalone highpass you can use however you like. There’s more to follow (I did mention the aliasing thing designed around hard clipping) but RetroBass leads the way. :)

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.

Meter 0.4

https://github.com/airwindows/Meter/releases

OK, buckle in.

Meter, as in Hit Record Measuring Meter, now is loudness agnostic.

That’s right! I found a way to map length of clippage to the peak readings and make it continue to give readings as if peaks were present. Just as it’ll want the peak levels to form a ‘cloud’, it will want the clip lengths to be unpredictable, but you can get away with silly amounts of clipping before Meter stops being able to evaluate it. It’ll more or less accurately determine tone colors and let you make adjustments completely beyond loudness war thresholds, all that will happen is the dialing-in process gets more twitchy and hard to control the harder you push it. And that’s just reality so there’s no ‘fixing’ that, back off on the gain if you want to get sounds more easily :)

Now, there’s a very long video that shows Meter responding to many different real hit records (censored with ContentHideD) so you can study that to identify the different patterns and tone colors to pay attention to, and you can just use it and learn how it reacts, and if you don’t like lots of reading then go and experiment. Lots of text will follow. It began in the source code as comments, and it explains exactly what Meter does, so if you’ve not TL;DRed by the next paragraph, it’s time for reading. (over 2000 words explaining everything)

Power is the intensity of varying peak energy, between maximum and minimum. It can’t be always maximum because that is just constant loudness and can’t startle or vary. Power is the derivative, the unexpected. it’s the ability of sound to be producing peak ‘aura’ beyond what our ears think is the median loud. We do not leave crest factor to be quiet with it, we leave crest factor so we can constantly have the ear tickled by stuff happening, that alerts the brain to peaks far beyond what the loudness seems to be. If they aren’t present, the sound is boring. The key is DERIVATIVE: peaks matter to the extent that they extend beyond the RMS (root-mean-square, body/density of the sound),and power is the extent to which peaks are both happening and varying from each other constantly. That’s a second derivative. This cloud expands as the capacity for loudness surprise rises: if you have continuously playing sound, whether it’s a drone or saturation or simply a reverberation in your mix, it’ll restrict the ability of the peak cloud to drop lower. Limiting the volume or bass of such sustaining elements works to widen the peak cloud’s range, bearing in mind that the ideal situation is constant peak activity at every dB level at once at all times. This is of course impossible and a contradiction, which is what makes it interesting :) We hear ‘activity at very low dB during loud sounds’ as openness of the sound, drama/excitement, and mix ease of listening.

Power is drawn as a cloud of blue dots in the peak meter, and the bigger the blue dots, the more sonority, loudness, and power are happening.

Detail is the intensity of varying slew energy, between maximum and minimum. It can’t be always maximum because that is just hardness and glare and sounds bad to people. It also can’t be minimum or it comes off as dull and uneventful in the treble range, so there’s a balance to be struck. Recording mediums used to set hard limits to how much treble you could have in the very high frequencies, an artificial limit, but it accidentally served to restrict detail to where it would balance with power (maximized peak energy) because it would end up quieter, balanced with the more hearable RMS loudness, allowing the power to be experienced as a sound event. Modern technology has no such limitations, so detail is commonly made to balance with RMS loudness in a context where ONLY RMS remains and the peak energy is not there anymore. (also, detail can contribute to loudness) That means you can get a ‘correct’ RMS/detail balance, but it only highlights the lack of power from missing peak energy. When you balance detail with a loudness balance that retains power and uses it, it still sounds ‘correct’ but the aura of the thing is very different, because the active peak energy is also sensed and it too can be in balance with other parts of the sound. This is the same phenomenon as the balance of taste stimuli in Heinz ketchup. It’s like a cheat code.

Detail is drawn as a cloud of red dots in the slew meter, and the bigger the red dots, the more detail and high frequency stuff is happening.

Authority is the amount of varying zero cross energy, between maximum and minimum. It can’t be always maximum because that is just one frequency and reads as boring, plus it doesn’t translate across bassbins as well. The zero cross meter measures how long the audio can go before crossing the middle of the waveform again. As such, loudness of bass makes it go higher and loudness of treble tends to disrupt this and take away low zero cross data. Thing is, high and mid frequencies do also show up here, and it is just as subject to the need for an ‘evenly spaced dot cloud’as everything else, but we hear evenness of this cloud as a sense of scale, especially when it reaches low frequencies. Simple resonant lows down here produce only one point of zero cross, which is why something as simple as an electronic kick that drops in frequency ‘translates’ better and is more interesting to feel and hear. Sweeping across bassbin resonances works better. But the way higher frequency content interferes with the zero cross is mix-dependent too: for instance, if you lack low mids that should be there, the zero cross cloud will have gaps of no data and then the bass all by itself all the way at the bottom. Bringing in audio at other frequency ranges directly brings in zero cross ranges as they interfere with the zero crossings, and this can be tracked as a dot cloud like the other meters, but maximizing this area (especially with the inclusion of DEEP bass) is what produces a sense of imposing, majestic authority to the track. This is pretty easily screwed up by having RMS too high, peaks too clipped, and detail boosted to match what RMS is doing. The funny thing is that, since this can be made to balance with peak energy and not simply RMS, that means if you have a high power mix with lots of energy beyond simple RMS, maximizing authority means hitting on all frequencies with perfect low mids. That is a recipe for bass authority, not simply bass loudness, so authority is ‘power through peaks, but for the lows’. Considering what a black art that sort of thing is, it’s useful to view it in this light.

Authority is drawn as a cloud of green dots in the zero cross meter, except that bright green only happens with quiet bass. As the bass gets louder, it’ll draw darker and darker greens and larger dots, so the dots in the zero cross meter are bigger and blacker when you have more bass impact.

All the colors can apply to all the meters, they’re just sized and positioned differently to suit the needs of the meter. Zero Cross in particular can have a lot of information to take in. Generally you want to be able to pick out important sonic events as big dots of whatever kind, and identify them visually as well as hearing them. If something important is happening but it doesn’t show up in any meter visually, it’s probably worth fixing the sound so it stands out more. That’s a risk of loudenating: you want to be able to pick out musical events and not have them wash out.

The Hit Record Letter Grade is ‘totalPackage’ followed by Power, Detail and Authority in order. That’s what’s in the video, but be warned that I may have to alter this order. I’m underway studying huge numbers of existing hit records and measuring them again (keep starting over) and have observed something important: of the legitimate top 10 hit records, they’re almost always having the Authority letter end up closer to ‘A’ than the Power letter. I make a little four letter word so I can sort things alphabetically, and I’ll need to alter the order to go letter grade, then Authority, then Power, then Detail. So be warned, the functionality of the meters won’t change but I’m moving the bottom one to the top since the video was made :)

The importance of the letter grade is that it’s a running tally of how well balanced each of these existing categories are with each other. It doesn’t work like Power/Detail/Authority which are tracking the highest score in each of those categories and can only increase (doubleclick the meter to reset everything). Meter is taking the information from those dot clouds, and producing the tone color that’s drawn in the color bars (at tiny sizes, it draws the whole meter in that color, with a gradient). The goal for achieving balance is to get each part of the color, blue for power and red for detail and green for authority, to balance. They’re driven by how broad and dense and even each cloud of dots is (see the video for examples). The letter grade will be swayed by louder stuff more, but does still pay attention to intros and fades, based on how loud they are. This overall balance measures the WHOLE song, so it can go up or down if the focus of the song drifts. To start over, doubleclick the meter to reset it, and you start measuring from scratch again.

You can sometimes see cool moments in a song be represented by the tone color swinging to white when something happens: get to the good part and it turns out to show perfect balance.

The color bars drawing tone color at any given moment, also directly produce Airwindows Meter DIRECTIONS. Those instructions come directly from whatever tone color is in play. You’ll find them in the second tone color bar between meters: no AI of any kind is used.

These are the messages you can get, depending on how strong the tone color is (perfect balance is a white color and there’s no directions when you’re really close to perfect, these are when adjustment is needed)

‘less bright’ simply means, there’s too much red dots and brightness, tone that down. Other aspects like fullness and loudness are good. You probably don’t have to turn brightness down by much, tiny little adjustments will do it. If you’re a brightness fan, you might want to see always tiny amounts of this message, just don’t get carried away.

‘too much air’ is a combination of that and good zero cross behavior, meaning your whole mix is super flashy and huge. Think The Fixx or a Trevor Horn mix. The phrase is from a Frank Zappa album cover :) It’ll sound good, but you can let things be a bit more relaxed and a larger audience should enjoy it.

‘allow fullness’ means zero cross is overdelivering and other aspects aren’t keeping up. Think ‘Money For Nothing’ or ‘Purple Rain’, this means it’s a spectacular mix but the foundation is lacking, so allow a little more bigness to it.

‘allow detail’ is halfway between loudness and zero cross. Mellow or ambient stuff might well see this. It can mean turn up the treble, or simply have brighter events happening, which would be an arrangement thing.

‘allow activity’ is only loud, but both slews and zero crosses are either absent, or so limited to one narrow zone that the sections don’t register. Think drone music, very static things, and so it’s saying ‘let more things happen maybe, if you want to get a hit record’. Or you can intentionally target these responses, either gently or maxing them out.

‘less boring’ is the above, but stuff gets more intense! Right up to the edge of ‘modern metal wall of ultra-bright’. If you’re trying to be brutal but you get ‘less boring please’ then consider whether you’re making more of a drone-music than you intended. Meter will tolerate loudenation much better now, but this is a risk you run that you won’t get if you leave peak space for impact and drama. If loudness is the only guideline, Meter may find it boring when it doesn’t vary at all.

‘less harrowing’ is where Dillinger Escape Plan gets to sometimes. Extreme heavy music that’s also max loudness can go beyond ‘loud’ to where the brightness just savages you, and this is the tone color that produces. If you mean to be harrowing, it’ll cost you in hit record potential, but this is a tone color you can target if you like. Dial it back for a broader audience reach.

‘more calm’ means everything’s very bright and loud and impactful. Think ‘The Fixx and Trevor Horn’ again, but at their most aggressive. The idea is, this is a sensory overload that leans towards detail more than just a savage attack of sound, but it’s still hyped up an awful lot. It wraps around to ‘less bright’ again, but in this case it’s not just about ‘slightly darker’, there’s also an implication the loudness has got involved and your peak energy combined with the brightness make the listening experience more dramatic than you need. Some genres do well to err in this direction.

Older Posts

Airwindows

human-made bespoke digital audio

Kinds Of Things

The Last Year

Patreon Promo Club

altruistmusic.com

Dave Robertson and the Kiss List

Decibelia Nix

Gamma1734

|GRENS|

GuitarTraveller

ivosight.com – courtesy Johnny Wishoff

Podigy Podcast Editing Service

Super Synthesis Eurorack Modules

Very Rich Bandcamp

If you’re pledging the equivalent of three or more plugins per year, I’ll happily link you on the sidebar, including a link to your music or project! Message me to ask.