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.

BeziComp

TL;DW: BeziComp is a radical compressor that eliminates aliasing!

BeziComp in Airwindows Consolidated under ‘Dynamics’ (CLAP, AU, VST3, LV2)
BeziComp.zip (706k) standalone(AU, VST2)

Funny how many of my Bezier curve experiments make plugins with powerful weird colorations… and then, this one is the opposite.

Or at least, appears to be the opposite… until you dig a little deeper, and unleash the madness.

So this is an experiment. The full range of the experiment is available, meaning ‘good’ settings might occupy small and fussy settings on the controls. It’s Bezier curves as the result of literal sound density, so it’s a compressor without attack or release, just a curve that goes wild based on what’s under it, audio-wise. There are three controls that can either harness this, or just turn it loose to be weird and unexpected.

Comp drives the Bezier curve: it acts like your threshold. Everything you do ends up interacting not so much with ‘compressed level’, as if you can set the threshold and it’ll be smoothly compressed below it… no, BeziComp works largely with HOW you hit this threshold and interact with it, because if you crank Comp way up it will wrap around not to a flat amplitude, but silence. That’s intentional, because the output is meant to work with what you might call a ‘live’ curve rather than flatten things into a featureless dynamic line. So there’s a start: Comp is how much BeziComp reacts, but it flattens out to silence, not audio, so you keep Comp in check (unless you are specifically just isolating attacks in Wet using this behavior).

Speed is basically DeRez, except the range goes way deeper into subsonic frequencies than usual. It still goes way up into the audio band if you like. The key here is, this sets the energy level of the Bezier curve based on a loudness window that’s directly related to the Bezier curve itself. So in theory, it would smooth things out completely, except we often don’t use compressors for that, do we? We slow attacks and speed releases to get sonic effects. Turns out that’s what happens here, because as BeziComp reacts, the reaction is slowed by having to analyse the audio, and so as you slow it more you get a broader and broader attack on the sound. If Speed is high, it’s real twitchy and will jump on transients quickly, but if Speed is real slow, you have a slower ‘swing’ that can be timed to a beat and used to accentuate the groove. So far, basically normal (ish).

Dry/Wet is basically your ratio control, but extra. Since BeziComp wraps around to dynamically invert, Dry/Wet is the only way you can get continued sound if you’re pushing Comp real hard. Anything over about halfway gets you into territories that act like vari-mu tones: the ‘squish’ abilities here are very extreme. Even when keeping Comp and Speed in check, when using BeziComp on something like mix buss, it will probably still be almost all Dry, because that’s the only way you can force it to have a relatively low ratio. Expect to not use full Wet in many cases, treat it like full wet is sort of ‘isolated delta of the effect’, a more exaggerated version of what you want.

Now, here’s where things start to go off the rails a bit.

BeziComp is modulating a Bezier curve, not following an attack and release. Speed does profoundly affect this, but not in the sense of setting a maximum speed for the behavior, instead it fixes a tightest corner with which the curve can TURN… and it’s constantly willing to use that sharpest corner, and it will apply that corner to anything.

That means if you have full silence followed by full density, BeziComp will attack harder and compress more than if you have just as hot a peak, but less audio behind it. It is NOT a limiter, or even a normal compressor, If you have loud audio and then sudden silence, it will begin swooping up in loudness not instantly, but on the same curve (and minimum curve radius) and then it’s gonna put another curve radius on there as it hits silence, rather than simply ‘switching’ to full volume.

This means BeziComp is more free from aliasing than any other compressor, period, even at high Speed settings. Nothing you do can make it suddenly hit the threshold and start to turn down. The amplitude modulation IS the Bezier curve, meaning it can only contain harmonics below what you set, meaning no matter what you do it can’t produce an artifact over its own curve radius. And at low Speed settings, that radius is VERY wide.

So, BeziComp is both able to make unexpected moves (since its maximum gain-change speed is not an Attack or Release, but whatever its Bezier curve allows it to do) and also hard to hear (because you can’t go by artifacts, there are none). It’s disgustingly transparent but also capable of being quirky and throwing odd bursts of loudness or silence in there. No matter how extreme you make it act, it hides the extremeness through using the Bezier curve on dynamic modulation… and no matter how well it hides its moves, it’s still capable of unexpected quirks, because of that fact that the attack and release speeds aren’t really just ‘speeds’ but curve radiuses.

We’re not used to using compressors that do that. Time we learned, because I think it’ll be good :)

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
download LinuxARMVSTs.zip for the 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.

SoftClock

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

SoftClock in Airwindows Consolidated under ‘Utility’ (CLAP, AU, VST3, LV2)
SoftClock.zip (512k) standalone(AU, VST2)

By special request, here’s SoftClock before I can properly demo and use it. I’ll have to play catch-up!

SoftClock is a timing reference, like a click track… but far more powerful, because it does not synchronize machines so much as it synchronizes HUMANS. You put it on an unused track, and route it to another track AS AUDIO which you then record, in the arrange window as just another audio file. This can be done as you lay down an initial track while listening to the SoftClock output.

What’s the noise it makes? Frequency swoops, of a tone. It doesn’t turn on and off, there’s no attack or decay… a continuous tone which drops in volume as the pitch drops, but otherwise is just a drone… and this is your new click track.

The next question is ‘for the love of God, why would you do such a thing?’ Because not only is this not a percussive sound, it doesn’t even have a definable attack, and yet it proposes to replace click tracks, and in fact wants to become a timing reference for a world of ‘human played’ music where nothing is snapped to grid. There will be no grid, just human-played tracks riffing over SoftClock.

The reason is, Entrainment. The ‘swoop’ of the tone mimicks the swing of a drummer’s arm, or the strum of a guitar… or a dancer’s body. It’s digitally accurate, but directs the listener not toward an inhumanly accurate ‘tick’ but towards a motion… and when many players are moving in synchrony, that’s when you have a powerful groove.

But there’s more. It’s also a reference to where you are in the bar. The non-accented beats ebb down to a lower tone, and then ramp up again toward the One, which is the loudest and highest-pitched swoop of the tone. The amount of variation can be controlled, as can the basic pitch. Bear in mind the volume does modulate, but not in such a way as to produce an attack, even at the fastest speed settings. You dial it in to mimic the motion of your body playing, whether that’s groovy or laid-back or hammering, and the non-accented beats always show you where you are in the bar, in an easy-to-hear way, even if the beats are themselves swung or the backbeat is slowed to add weight and bombast.

What about those accented beats? Firstly, they’re louder and higher-pitched, but they’re also spaced out in an interesting way. You see, SoftClock can give you 4/4, or eighth notes, or four bars of 4/4… but it will also give you odd times beyond your humble sevens and elevens. It’ll give variations on funny times when you’ve gone beyond 4/4. Here’s the list, and how they’re counted:

1, 2, and 3 are counted as you’d expect. 4 is counted as 2,2. 5 is counted as 3,2 (in other words: one two three one two, with one being the accent). 6 is 3-3, and 7 is counted backwards from how Pink Floyd’s ‘Money’ had it: 4-3, one two three four one two three. 8 is 4-4, 9 is 3-3-3, 10 is 5-5.

Then things get more complicated. Note that Count numbers 8, 16, 24 and 32 are reserved for 4/4 with accents every eight beats, so you can have normal time but hear the unaccented beats guide you around one, two, three or four bars and highlight the first One.

11 is counted four ways: 3332, 443, 551, 65.
13 is counted three ways: 3334, 445, 553.
17 is counted four ways: 44441, 5552, 773, 881.
19 is counted four ways: 44443, 5554, 775, 883.
23 is counted four ways: 444443, 5558, 7772, 887.

Count the accents and they’ll come in these spacings. It’s options for ‘sub-phrases’, repetitions or modulations up front and a variation for a turn-around, orienting you to the odd rhythm. The idea is for the repetitions to themselves be easily countable, and ideally seem to fit into a ‘four sections but one is different’ or ‘three sections but one is different’ model.

This is in line with a music arrangement game I’ve been developing that’s more likely to see life as a video-game now than as the card game I’d intended. Point being, SoftClock can do normal metronome things (except using entrainment rather than ticking at you) and orientation things as far as locating you in the bar, but it can also combine all that into a baffling proggy rhythm that nevertheless guides you into grooving and memorable riffs of many kinds. If the standard variations on freaky prime-numbered prog-meters aren’t enough, you can play with Swing and the delay of BigBeat (accented beats) until you’ve gone full Beefheart: or, rather, full Magic Band and John French drum parts.

Dive as deep as you like, or keep it more simple: SoftClock has you covered. You may be surprised at how much easier it is to hear, entrain, and orient yourself when using a click track that refuses to click. Remember, lay it down as a track, don’t try to match it to a normal click, don’t grid anything! SoftClock is for gridless grooves, and music that sounds human because it moves like humans. See how it feels.

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
download LinuxARMVSTs.zip for the 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.

PointyGuitar

TL;DW: PointyGuitar is a supremely adaptable instrument amp.

PointyGuitar in Airwindows Consolidated under ‘Amp Sims’ (CLAP, AU, VST3, LV2)
PointyGuitar.zip (531k) standalone(AU, VST2)

Sometimes the best way to get the performance of a very analog thing in a very digital format is to accentuate, not the spirit of the analog, but the capacities of the digital. What you get then is not fake gutless imitations of the artifacts of the analog thing, but the heart of why we go for that analog thing. Meet PointyGuitar, which is all amps. Just adjust… if, that is, you know what the analog sound is made of.

Here’s how it works. SmoothEQ was a sort of breakthrough: the ability to get very steep EQ crossovers while still zeroing out at perfect, unaltered fidelity. It sounds the way it does because it reconstitutes the original source between every stage, rather than doing independent EQ bands and trusting that they’ll be ‘just as good’ because the gains and EQ slopes match up closely enough.

Then there’s AngleEQ, a sort of disaster: a very heavily colored EQ where the crossovers are so phasey that it’s like running stuff through a speaker. The opposite: whether as a highpass or a lowpass, it produces a weird physicality and scrambles the sound pretty good.

PointyGuitar is both, in series, with a basic distortion (as found in FireAmp) between every EQ stage. There can be lots of these EQ stages, so there can be lots of stages of this simple softclip, but it’s always on the full sound, not a ‘multiband’.

What happens? Any given section might add small amounts of overdrive. It’s not an electrical circuit and it’s run at double precision so there is no thermal noise/Johnson noise: it’s just a bit of overdrive if you’re boosting a band. Otherwise, if it’s flat it’s extremely flat (interestingly, inside the flat EQ it’s still using the AngleEQ math, but in such a way that it sums to perfectly flat. Using biquads would’ve got steeper crossovers, but requiring more stages turned out perfect here)

As it’s passed on to the next stage, any distortion harmonic that doesn’t fit into the following band just plain gets filtered out. This is most notable in the amount you can crank the ‘High’ band. It’s around 3K for guitar chug, but it’s resisting aliasing super hard considering the amount of gain you can add. Same applies for every band except Presnce. You get that, High, Mid, Low and Sub. For high gain ‘5150’ type sounds, you boost High and cut Low.

Once you have your sound, be it super high gain or clean Fendery stuff (again, boosting High often helps things sound more amp-like) it’s over to the highpass and lowpass. HSpeakr cuts off the bright, and LSpeakr limits the size of your virtual cabinet. They’re pure AngleEQ and combine to produce a bandlimited sound without use of a cabinet impulse, harnessing the intensity of the EQ/saturation stages so that the tone sits right. Very delicate adjustments are what you want here, but both those controls have pretty much full range adjustability. 50 foot speaker, or clock radio, are available if you should need them.

In this way, first building a sound out of identifying which frequency bands need most saturation and which to back off, and then channeling that very saturated but very clean sound through extremely colorful bandpassing, you produce an amplike sound (guitar or bass, any amp of any kind, it just depends how you adjust it) which fits immediately into whatever space in the mix you need, with the right tone colorings…

…at ZERO LATENCY.

It runs without oversampling, and all the EQs operate directly without pre-ring, as analog circuits do. So you can track directly with PointyGuitar, set any way you like, dial in the basic sound of any sort of amp no matter how clean or dirty, and have it respond so immediately that you’ll notice if anything else in your recording chain is adding latency. It’ll feed back like any real amp, it’ll feel connected to your fingers, and you can dial it to do anything you want, pretty much. If there’s a cabinet honk or something that you actually want to add (rather than just remove) you can run the whole thing into ConsoleX and dial that on the parametric, perhaps in conjunction with the Speakr controls.

I may have just replaced my little tube amp and iso cab and new speaker DI box at a stroke. Didn’t expect that, but both my ears and Airwindows Meter tell me PointyGuitar is in the zone, even for really difficult sought-after guitar tones. The range is pretty shocking: it ought to do bass amps, Plexis on Variacs, ratty little Peaveys, you name it.

Oh, and you get a gate (basically DigitalBlack) that triggers off the input before the distortion, but gates between the ‘amp’ and Speakr so that even if you’ve got it firing really staccato and quick, it’ll merge with your string-damping and act like part of your playing, for guitar OR bass.

Enjoy. This should work both for vintage tone guys AND Reznorian madmen. :)

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
download LinuxARMVSTs.zip for the 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.

SmoothEQ

TL;DW: SmoothEQ is a sharp, accurate, transparent three-band filter.

SmoothEQ in Airwindows Consolidated under ‘Filter’ (CLAP, AU, VST3, LV2)
SmoothEQ.zip (526k) standalone(AU, VST2)

This follows AngleEQ, so it follows an extremely weird, colorful, poorly behaved EQ with a super-clean, accurate, well-behaved EQ, just for variety. Not in the sense that you dial in exact dBs and frequencies: it’s still a ‘by ear’ filter like the great analog EQs, and I’m probably not the person you go to for ‘add exactly two and a half dB of 3026 hz shelving and that’s the important part, all EQs sound the same’.

When I say accurate, I mean that unlike AngleEQ if you set SmoothEQ flat you get EXACTLY what you put into it. It uses a subtractive technique that I like, not sure if other people do this, where the sound is assembled out of sub-bands that are always ‘one band is the filtered, and the other is that subtracted from dry’. It’s that which I couldn’t do with AngleEQ, and which has been tricky with everything short of simple biquads (for which it works excellently).

And this is the breakthrough where I worked out how to do this thing I like, which I’m not sure anyone cares about but me, with ANY degree or order of filters so long as they’re biquads. Maybe even beyond that, but turns out biquad filters are very cooperative with this.

SmoothEQ uses eighth-order crossovers: steep! They isolate real well. But so powerful is the technique that I can use Bessel filters instead of Butterworth, for nicer phase behavior. There are limits: I don’t think I can use this for speaker crossovers, it’s strictly for setting relative volumes of EQ bands and listening to a combined output. The trick is, if I apply cascaded filter crossovers and try to subtract it from dry, I get what I got in AngleEQ, a phasey mess.

But if I reconstitute the original signal from the ‘sections’ between EVERY stage of filtering, I get the degree of filter steepness I wanted, AND all the sums still reconstitute to a perfect, bit-identical input if set flat.

If you wanted an Airwindows five-band, or seven-band, or 31-band EQ this is how I’d have to go about it. I don’t know if it would hold up at 31 bands of EQ, or how well that would work, but mathematically I know that if you set it flat you’d have ultimate, bit-identical, perfectly transparent sonics, because that’s the only way the technique CAN work. There are obvious applications in making, for instance, a Mesa Boogie 5-band guitar EQ and having it sound amazing without analog modeling: sometimes what you want is the merits of analog processing, not just to imitate everything. It’ll make EQs that are perfectly accurate set flat, and then you’re shelving up and down bands of frequencies with extremely clean boundaries, as steep as you like.

SmoothEQ is a simple three-band version of this. Now that I know I can do it, I can do a future ConsoleO (orchestral) and really do a good job of keeping the tone even through filtering. This is not a linear phase EQ but does not need to be, since everything is either perfect or some degree of the tone produced by extreme EQ isolations: whatever that sound, if you add only a tiny amount of filtering you automatically get only a tiny amount of that tone color. That’s why I build it in that peculiar way, to get that result.

It’s going to be fun running with this technique and making stuff sound better, and I hope you like SmoothEQ :)

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
download LinuxARMVSTs.zip for the 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.

Older Posts

Airwindows

handsewn bespoke digital audio

Kinds Of Things

The Last Year

Patreon Promo Club

altruistmusic.com

Dave Robertson and the Kiss List

Decibelia Nix

Gamma1734

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.