Airwindows

Patreon 2: More Stuff More Often

So there are more than 150 Airwindows plugins out there, free and open source in Mac AU, Mac VST, Windows VST and Linux VST formats.

More than two years ago, on August 3rd 2016, I announced my Patreon, and my goals. The idea was if I earned enough money that I could continue doing Airwindows even after the loss of Kagi, my previous payment processor, I would begin releasing all the plugins I formerly sold for $50 each and all the ones that were free as well. If I met a higher goal, I would begin open-sourcing the plugins. If I met still higher goals, I’d begin doing it faster, do more of them per month, accelerate the pace until people had all their favorite Airwindows plugins for free with open source.

That was more than two years ago. It worked. I still do plugins full-time, I’m nearly done with the backlog of Kagi era plugins (for instance, Compresaturator is coming tomorrow) and everything is open source now as soon as it’s out. We won: we got everything to work out in the best way, and though I keep doing new work there’s not much left out of the ‘goodie bin’, and many bases are covered over and over again.

This means it’s simply not as good of a goal as it used to be, to say ‘I will release the plugins, I will open source them’. I’m already doing that. People are used to it, perhaps even take it for granted. And I’m still struggling to live, kind of: I’m very limited by my economic position and could do more (and be healthier and less scared) with more stability.

So the question becomes, what MORE can I do, for you all? Because there’s no sense saying ‘look at all I am doing’, if I can think of something new and different and more interesting to do.

The answer lies in the livestreams I’ve been doing on Mondays 11:00 AM EST, and in my own past projects.

Back in the day, I did an elaborate study of evergreen hit records, trying to work out the secrets to why people compulsively love certain records and buy them in the millions. It didn’t directly help me sell plugins back when I had to do that every month to live, so the study went away, and the follow-up examining records in more detail never saw daylight. I had to work and sell plugins, but some people remembered it. Well, one of those people is a writer, and has recently written an article in the New York Times about the loudness war and what it’s done to music. It’s a lot like my study. That’s because it was based on my study from word one, and written in consultation with me, and I’m one of two people cited in it: the other is Bob Ludwig, noted mastering engineer. (technically, that alone might make me notable enough for Wikipedia, but that’s not for me to meddle in)

So here’s the new goals.

If I can earn $1500 a month, from everybody combined, I will add another livestream to my schedule, and spend more time with you. It’ll be Wednedsays 11:00 AM EST, and I will spend two hours analysing and explaining a ‘hundred hits’ evergreen classic album, using the measurement techniques shown in the video. These don’t exist in plugins, and you can’t get them elsewhere. That metering was done before I did plugins, and won’t run in any DAW. I’ll talk about everything I know, regarding the album we’re analysing: two solid hours of tips and tricks and research from every conceivable angle. And then, the next Wednesday, I’ll do it all over again with another album.

If I can earn $2000 a month which probably won’t happen right away, but that’s what goals are about… I will continue this, but I’ll begin including one minute clips from my own patrons who are supporting the Patreon at $50 a year or more. It’ll be the same two hour show, but I’ll be devoting my attention to your music. Kind of like a production session: bring your fans and treat it like a showcase, because when I produce my goal is to find what’s amazing and desirable about your work, and dive into that with encouragement and enhancement. I’ll tell you stuff about your music that even you might not know, and will guide you towards music that is a better you. That starts happening at $2000 a month.

Oh, one more thing: since I’m using one minute clips and will be talking over them some of the time, I am prepared to play audio from the classic vinyl that created these evergreen cultural artifacts. These studies go back to the source. This is what got people excited about this stuff in the first place, and much of it will come across even over an internet stream. It’s exactly what I’m trying to teach people to do in their own music production.

I’ll be doing this live and there may be instances where the record companies nuke the saved livestreams and threaten my channel, because they do not care about fair use at all, so tune in to the actual livestreams if you’re in doubt. If all else fails I can talk about the measurements without playing audio… but I intend to let you hear what this stuff sounds like, though you won’t get the whole song (just the one minute as shown on the chart).

There can be more goals beyond that, but no sense being too concerned with them until they’re in striking distance. The point is, I’m doing a free plugin that is open source every week, and pushing the boundaries of digital audio, for less than $1500 a month. To earn more, I need to do more. I can’t put out more plugins because the market can’t even absorb what I produce right now, so I need to spend my time actively engaging with people and helping the struggling musician: bringing back the Hundred Hits study and then opening it up to patron content makes sense, and it’s something I can actively do to earn that higher sum. I’ll continue to try to think of more than I can do, and since I understand audio things for a living, Hundred Hits is an obvious choice to start with.

Join me and reach that goal, and we’ll learn how to make sounds that sell millions… or at the very least, learn a lot more about the sounds that HAVE sold millions!

Date March 16, 2019
Author Chris
Comments 1 Comment

BlockParty

TL;DW: BlockParty is like a moderately saturated analog limiter.

BlockParty.zip(359k)

What do you get when you cross an Airwindows compressor, with OneCornerClip?

Pretty much this. Okay, so it wasn’t a very difficult riddle, was it?

BlockParty acts like a somewhat distorty limiter. It’s not at all about lookahead (in fact it doesn’t have any) or preserving tones pristinely. Instead, it takes the onset of sounds that would compress, and manipulates the attack in the way that OneCornerClip does. The threshold gets kicked way down, and gradually expands to full scale, and since the threshold’s determining compression, that means BlockParty doesn’t have a stable compression threshold. It’s interactive with the audio you’re giving it.

Because it’s on the OneCornerClip model, that means it’s a mostly-compressor with OneCornerClip-like behaviors. That means bass which blooms and has fullness even under heavy load, and highs that don’t poke out or distract. The result is a thing that sounds real analog-y but not super clean. You can use very small amounts of it (there’s a lot of gain on tap) to do peak limiting for loudness maximizing, or you can slam things into it for effect. It’s called BlockParty because heavily limited stuff sounds like blocks of loudness: it’ll get you some of those sounds, but not as cleanly as your classic ‘loudness war’ limiters. It’ll also smash drums and things in its own distinctive way, which might be its strongest suit. On the end of it is a clipping stage to make sure nothing you do will ever produce overs. The clipping stage is AFTER the dry/wet, so to get a true dry you’ve got to turn it off: this is because raw digital clipping is another style of loudenating, so if you were going super-hot into BlockParty and wanted to dial in some pure digital clipping you could use the dry/wet to do it (or, if including some dry would have given you overs because your direct buss signal includes overs).

BlockParty is a fierce loudenator with a voice and style all its own, using techniques that are distinctly Airwindows. It might be just what you needed, or it might be a little too grungy for you… but either way, there’s nothing quite like it, so check it out. It is open source and completely free, because I get paid through Patreon. The idea is that if you would’ve bought this at $50 for perpetual license on all your computers for life plus you get the source (heck of a deal for $50 really) then you should jump on the Patreon if you can, and add $50 a year to whatever you’re patreoning. That keeps me developing, and the only way you get new Airwindows developments is if I’m still developing. (stay tuned for some more goals I’ve got: since I am already doing open source and plugins as fast as the world can stand to see them, I must find new ways of giving more time and attention on top of that if I want the Patreon to get bigger faster. I’ve got some plans that are worth being goals, and that’s next on the agenda besides the plugins I’m already doing.)

Date March 10, 2019
Author Chris
Comments 3 Comments

PurestSquish

TL;DW: PurestSquish is an open-sounding compressor with bass bloom.

PurestSquish.zip(355K)

This time, the video’s enormously long so I’ll be brief: PurestSquish is a compressor, with its own sound. In the video I compare it to Pressure4, Logical4, and SurgeTide, and also show how it can be used in conjunction with SurgeTide (a real ‘sleeper’ plugin not easily understood) to produce amazingly transparent dynamics control.

I also spend some time torturing it with sine sweeps, showing how Pressure and Logical are more like ‘analog emulations’ and produce harmonics, while PurestSquish instead does a weird thing when you turn off and on signal generators. So if you’re looking for ways to say ‘this is broken forever!’ watch those parts of the video :)

If this doesn’t worry you, PurestSquish also has a bass bloom control that lets you pass subsonics or bass notes through uncompressed, to taste. If it does worry you, chalk it up to PurestSquish running simultaneous two-and-three-sample-interleaved compressors, much like Capacitor runs two-and-three-sample-interleaved filters, and use one of the other compressors I’ve put out, perhaps one of the three also featured in this video.

Patreon is the reason I’m able to do stuff like this, including this weird trick of releasing odd plugin designs that other people don’t let you have (for very legitimate reasons). I wouldn’t be able to put out PurestSquish as it is, under strictly commercial conditions: I’d have to sand the rough edges off, and that might ruin it (much like the bugfixes for the original Console5 fixed the tendency to DC offset at the cost of its special tone, resulting in re-issuing of the rawest original version later). In this case, PurestSquish ought to be good under most conditions, but it’s one of those plugins where the magic tone comes at a cost, so if it starts throwing 15K artifacts at you, consider using one of the other compressors for that sound. And don’t feed it test tones after midnight :)

Date March 3, 2019
Author Chris
Comments 3 Comments

MV

TL;DW: MV is a dual-mono reverb based on BitShiftGain and the old Midiverbs.

MV.zip(397k)

Back in the days of really old school digital reverbs, there were a couple weird and obscure ones that had a special mojo. I’ve got one: the original Alesis Midiverb. It’s quite low-fi and only has RCA jacks, but there’s a certain something about its sound.

Turns out one of its secrets isn’t so secret: the first two versions of the Midiverb don’t have a multiply unit. That means you can’t do certain reverb things correctly. Reverbs use a kind of delay effect called an allpass filter, which involves multiplying by 0.618 (I’ve sometimes generalized this to ‘the golden ratio to N decimal places’, where N equals ‘a lot’). But the old Midiverb couldn’t do that… so it made an ‘allpass filter’ by multiplying its stuff by 0.5. A bit shift.

Airwindows fans will know that there’s something special about a bit shift: especially in floating point, you can change volumes by 6dB pretty much losslessly. No, make that ‘totally losslessly’ since in floating point you’re only changing the exponent and could change it right back and lose absolutely nothing: the mantissa is never touched.

What would happen if you took this old school way of doing allpasses, and made a modern reverb out of it, using full-quality floating point to do it? What if you followed up by making the regeneration also strictly ‘bit shift’, increments of 6dB or infinite regeneration, losslessly? What if you added a way to roll off highs by averaging output samples of the allpasses, and did THAT entirely using bit shifts as well? And allowed for a big number of allpasses (26, all different increasing prime lengths), and gave varying treble rolloff by independently controlling which of the allpasses got the average treatment?

Welcome to the infinite land of MV. This is nothing like a normal reverb, but it’s got some great superpowers, not least of which is the ability to just sustain a ‘bloom’ forever. You can automate it by kicking the regeneration up to 1.0 any time you like.

You can dial in different degrees of highs roll-off using the bright control, or leave it at 100% shiny. Combining this with more restrained regenerations like 0.51 or 0.26 at medium-to-high sizes will give you very decent ‘impossibly huge reverbs’ of various characters. MV doesn’t do early reflections or plausible spaces, just the infinite wash, but that’s somewhat configurable.

It runs dual-mono, so you can dial down the size a bit (not too much or it’ll get nasty, you’re removing allpasses from the chain) and use it as an ambiance generator, and it’ll put all reverb tails ‘behind’ the sounds that make them: centered stuff stays centered, wide or stereo stuff goes super-wide. For this reason it’s very suited to use on auxes and submixes: you can add ‘space’ that’s very pure-sounding

It can do full, 1/2, 1/4, 1/8 and I think 1/16 level regenerations: set the feedback and it will use the bit shift amount nearest below the setting, so no matter what you do it will always retain its audio character. And the whole thing runs inside a PurestConsole instance except for the regeneration, which is extra… which means that if you build up a wall of infinite reverb, it can’t go into reverb runaway because distorted samples will wrap around and get quieter: you’ll have to trim down the output, but this makes infinite regeneration super-usable without applying any kind of compressor or limiter inside the loop. Since you can also do zero regeneration and it’s just a pile of allpasses, you can also do a ‘gated reverb’ effect if you like, which is good at airing up the mix but then getting out of the way.

It’s the first plugin to come out using the rigorous, exact-dither-amount version of my Floating Point Dithers. Everything else will follow, but it’s a huge amount of work so bear with me, and I’m also going to want to set up my new web hosting at least to put the next ‘NewUpdates.zip’ on, when it’s all ready. There will be big and awesome changes, such as being able to support all the bandwidth my customers need, and being able to co-locate and use load balancing. I’ve got expert help working on it, so when the new dither updates are out, you’ll be able to redownload the whole collection without hesitation.

When that DOES happen, if you hear a change in the dither (maybe just vaguely sense it, without being certain? If you’re honest?) it’ll be what MV has. The dither is now perfectly accurate, both for 32 bit and the seriously un-hearable 64 bit dither. It’s flat dither rather than highpassed TPDF, since the dither amplitude is designed to run at all different levels anyway. What that means is, the changes are very like the changes between standalone NJAD and the new NJAD included in the plugin StudioTan: it’s a bit darker, depths are blacker, there is less coloration of any kind, more neutrality. It also completely removes all truncation (not just the nastiest bits, but all of it anywhere) so we’ve gone from ‘thoroughly addressing a tiny problem that could build up’ to ‘completely obliterating the truncation to the point of mathematical infinity’. Oh, also it ought to run faster than the first version, because it’s using a thing called xorshift32 rather than the more high-CPU rand() function. So everything is better, including the CPU hit, and all that will be incorporated in the big update when it’s ready. Until then, all new plugins coming out will use the newest version. And DitherFloat is current. :)

Think of Patreon as my shopping cart. If you would have bought this if it was for sale from Kagi or wherever, at $50 for the plugin in all versions on as many computers as you like with free updates forever (how’s that for terms and conditions? That’s how the Kagi Airwindows plugins used to work, before all this), then please if you can, join the Patreon for $50 a year. If you’d have bought a bunch of plugins and you can do it without endangering or hurting your well-being, join for a bunch of $50ses a year: some folks are, and it’s very worthwhile as it turns out I can’t do it all on just mass quantities of $1 a month patrons, it takes all kinds to keep this show going.

Hope you like the show. There’s plenty more coming :)

Date February 24, 2019
Author Chris
Comments 8 Comments

Chris

Patreon 2: More Stuff More Often

BlockParty

PurestSquish

MV

handsewn bespoke digital audio

Kinds Of Things

The Last Year

Patreon Promo Club