Airwindows

Biquad2

TL;DW: Biquad2 is the Airwindows biquad filter that’s more sweepable and synthy.

This time, it’s time for the impossible!

As in, there was a reason nobody was doing sweepable, ‘synthy’ biquad filters…

Turns out it simply doesn’t work. One uses a different type, like state variable filters, for the synthy stuff. The reason is that, while biquads can sound pretty great (especially implemented like I do ’em), they fundamentally can’t cope with changing the filter coefficients mid-calculation. They flip out: you’ll hear some of that, especially at the frequency extremes. Low Q makes low frequency motions flip out, and high Q (and boy do I have a high Q for you this time) makes ultra-high frequency motions flip out in a really wild. glitchy way.

So obviously I gave up.

nope! Instead, I just kind of forced the filter into zones where it mostly is controllable. This is partly through REALLY smoothing the filter cutoff, especially at low Q and low frequencies. If you try and update biquad coefficients every sample (and I already changed the form from the more CPU-efficient to the more stable form: didn’t help much at all) the algorithm gets super twitchy, so part of what I’ve done is just stabilizing things. I tried for ages to come up with some bizarre hack to force the filter back into stability: no dice. So, the range has been limited a little, the Q doesn’t adjust below a Q of 1, and it reacts really slowly, because that was the only way I could get it to behave at all. (the original Biquad can sometimes be better behaved, because it’s only recalculating coefficients every new buffer. So, zipper noise. When you get rid of that your troubles get infinitely worse, with biquads)

I’ve also got the resonance (on low and highpass) working differently. I’m scaling back loudness, but I’m also applying a distortion, then averaging after that, to try and get an ‘overdriving filter effect’ not present in the purer original Biquad. You can go quieter into Biquad2 and avoid this, or not use as much resonance: I feel it’s kind of like the Roland Alpha Juno filter resonance distortion, not an everyday thing but when it pops up it has its own distinct quality that’s interesting.

Anyway here’s Biquad2 ;)

Patreon is how I’m able to stick around doing this stuff, and/or fixing up stuff that needed fixing. It’s become clear from the response to Monitoring (yes, I’ve got an update coming) that among the things that need fixing are ‘you make too many plugins!’ Since I can’t go and throw most of ’em away, I’ll need to do two further things: one, do more ‘combination’ plugins such as one that includes the MANY Airwindows distortion and saturation types. Two, come up with ‘sets’ of starter plugins meant to work together, which may include custom versions of plugins like Console, and three, get some more building blocks like basic EQs out there. Whoops, that’s more than two :) anyway, even if I can’t design a DAW directly, I ought to be able to tailor versions of Console and tone shapers and channel strip parts together and produce cohesive sets of plugins that are waaaay more approachable to the Airwindows neophyte. Since MV, I’d also like to do reverbs and really get a whole tapestry of mixing together.

We shall see. Enjoy the sweep-ier biquads :)

Date September 22, 2019
Author Chris
Comments 5 Comments

Channel7

TL;DW: Channel7 improves slew clipping and lets you blend Channel 6 and 5.

Channel7.zip(352k)

Short and sweet. You know how Interstage got an update because the slew clipping was too intense on high sample rates? This applied to Channel, too. So, Channel7 now handles high sample rates better than any previous Channel.

There’s more: Channel5 was the last one with the ‘Density’ algorithm, that’s extra fat-sounding. Channel6 got the Spiral algorithm, much purer and clearer… but without that fatness effect Density gets. My attempts to make a special algorithm that does both gave me Mojo and Dyno… neither of which are Channel or a replacement for either Spiral or Density.

So, Channel7 simply goes to 200%, not 100%. Up to 100% it’s the same as Channel6, but with the slew clipping fix. From 100% to 200% it does a crossfade into Channel5’s Density algorithm—so it ends up being best of all possible worlds. You can seamlessly go from perfectly clean, to the purest saturation there is, and beyond that to add any degree of fatness you like to the sound (and it fades between Spiral and Density, so there’s extra subtlety to the way it transitions).

Channel7: yeah, you’re pretty much going to want this upgrade no matter which Channel you liked. It is all of the best, more effective than ever.

If you ever jumped on the Patreon because of a version of Channel… you should take this for free ‘cos it’s an update to something you already have :D I guess if you never joined the Patreon, and you were impressed by Monitoring, and you never tried Channel, and all of it leads to you wanting to throw money, then go ahead and join (or I have a Ko-Fi as a backup plan you could use, or paypal or something). But this one’s Channel version 7. If you use Channel you should have this and not feel like I’m trying to squeeze more money out of you so definitely DON’T feel obligated to up your patronage just over this. Keep an eye out and I’ll try to do more that justifies more support. This one’s me giving back. You should get updates for free when I’m able to improve things a little more each time. Enjoy :)

Date September 15, 2019
Author Chris
Comments 5 Comments

Monitoring

TL;DW: Monitoring is your one-stop shop for final 2-buss work!

Monitoring.zip(412k)

…and then sometimes it’s a Big Deal. :)

Boom! Here’s your final plugin on the buss. This is one of those weeks where a lot of work comes together into a final shape. In a single plugin, you’ve got a pile of usefulness that you won’t have seen in one place before.

You’ve got Not Just Another Dither (24 AND 16 bit). The 24 bit is also used to make your DAC sound better on the other monitoring options (even if you have 32 bit DACs, it’ll still bring a subliminal naturalness from the Benford Realness calculations). You’ve got PeaksOnly back again, but also SlewOnly and SubsOnly: switch between them and if anything is either out of control or missing in action through these extreme monitoring situations, you know what to do.

You’ve got utilities like Mono and Side, for a quick mono check. If you split ’em to stems and then sum them, you ought to get the original signal back again so that’s a handy tool you’ll already be used to. But you’ve also got three new things, Vinyl, Aurat and Phone. This is where the ‘ultra-broad bandpass’ stuff went! It was too vanilla to work as a ‘Interstage’ killer, but the Airwindows biquad bandpass turns out to be ideally suited to test monitoring, where you want the bandwidth very limited but without sounding unusual or vibey about it. Vinyl just barely rolls off the extremes in the manner of traditional lathe electronics, which routinely discard stuff that CDs include. Aurat is a narrower band, like the output of those little one-driver mix check speakers: you can get a decent sense of that voicing, as if your speakers were made that way, without a lot of fake modeling or convolution to muddy things up. And Phone, well… Same, but if it was a cellphone.

But then you’ve got the crown jewel of Monitoring… Cans A and B. Airwindows Crossfeed for headphones! This isn’t an exact clone of anyone else’s product, this is MY take on what you need. It uses allpasses from PeaksOnly to blur and extend stereo content, first as a slightly delayed/smeared crossfeed, and then back again with a longer allpass to fill in some space. This does two things. It localizes stereo, but it also hints at the kind of information PeaksOnly gives you, so the Cans options both make listening easier, and mixing easier! Cans A is very subtle, much closer to the raw mix you’re working on, but with enough cues to orient you if you’re paying attention. Cans B is much more obvious, and is a good late-night check for if you’re getting fatigued: it’s a lot more blurry and ambient and narrowed, but it takes peak information that could get by you, and makes it obvious like a mini PeaksOnly. It feels like listening to the music off a stage, and presents everything you’ll need. And both of them use Console5 technology, Airwindows allpasses (as found in MV) and all gain operations inside them are done using BitShiftGain. So these monitoring options, whether subtle or plain, are set up to be the highest-fidelity monitoring options you’ll find.

So, it’s an exciting week, and I hope you like this omnibus monitoring/mix check/headphones tool/dither plugin. This is how you hear (if you’d like it to be). I owe it to Patreon, and if you’d buy this (perpetual license, lifetime support, you get the source code… you know, normal terms :D ) for $50 by all means go to my Patreon and kick in an extra $50 a year. The neat thing is, if you starve and have to drop off the Patreon, you get to still use the plugin! That is one of the reasons it’s better, and thus ironically deserves your subscription more :)

Thanks, everyone. If you’re half as excited about this as I am… then I’m still twice as excited as you :D

Date September 8, 2019
Author Chris
Comments 11 Comments

PeaksOnly

TL;DW: PeaksOnly is a transformative mix check tool.

PeaksOnly.zip(350k)

So I’m posting this super late, it’s a real all-nighter. Why?

Because this plugin exploded into my awareness pretty much just today and kicked everything else aside, demanding to be made right away and posted immediately. Not only that, I had someone email asking nicely if I could make a plugin that did for them what a busted subwoofer used to do for them: give some kind of cue if there’s excessive lows happening that’ll cause problems later.

All I could do was say ‘stay tuned’ and keep on working into the night to bring this one to light. (well, dark. It’s nearly 2 AM)

PeaksOnly is a mix check, like SubsOnly and SlewOnly before it, or Silhouette and other even stranger plugins. You pretty much need to not master through this one :D it totally wrecks the sound. BUT, it telegraphs information you might never have had, makes it more obvious than it’s ever been before.

Here’s how it works. PeaksOnly runs a little pile of allpass filters, but rather than just make them nice-sounding, it expands and expands and expands them, each time phase-rotating things a little more. It turns transient attacks of any kind into a little colorful wash, a flag of energy that stays at roughly the same level, but gets smeared out across time in a way you’d never normally hear.

Why does this matter? Because you wouldn’t hear it. Especially if you’re trying to work on headphones, brief sharp transients can become almost totally invisible, especially if they’re being peak-limited or get blended in with other sounds. That stuff can make your mix come alive, but how do you balance it? And if you’re on headphones, forget it: some peaks are just too brief, and unless you have a strangely powerful sense of energy levels that are normally invisible, there will be no managing the stuff that you simply can’t hear, the spiky intense sparks of audio like you get from passionate performances or tricks like using compression to spike up attack transients.

But with PeaksOnly, everything stays at exactly the same frequency balance (a trick of allpasses, especially mine which are prime-number spaced) but the bursts of energy, no matter how brief, get transformed into recognizable signals. If you’re short on transient impact, you’ll notice it. If you’re over-squashing attacks, you’ll end up with audio porridge. But if you’ve got a powerful, kicking, lively mix… every detail of it will be laid bare, turned into a sort of pantomime that exposes every transient at every frequency. Whether it’s how loud to make that snare or hi-hat, or how much sub-kick you can get away with, it’s all exposed. PeaksOnly is particularly fond of taking excess subsonic peaks and distorting ’em: you’ll never pack too much into the subs again.

It also tries to keep you at a sane loudness level (suitable for sending to mastering, if you really think you need to loudenate). Strangely, if you feed ‘mastered’ loud stuff into PeaksOnly and tolerate the distortion, it still reveals everything: there’s a difference between good ‘too loud’ and bad ‘too loud’, and PeaksOnly exposes that too. Anything out of balance, even in loudness land, will stick out.

I’ll be finding ways to further use this, but as always, you don’t have to wait. PeaksOnly (note the lack of controls! Remove or bypass it when you’re done checking) is there for you now. And now I’m gonna post it and go collapse into bed, ‘cos this has been a good day at the office :)

Patreon is how I do this. I don’t know anyone else who could’ve invented this one: it’s weird, sounds ugly, and yet it might be the best plugin you ever found. If you think so, please join my Patreon. If you do, you get to listen to my private little Evergreens podcast, a Fair Use exploration of how classic records got their sound (and how you can too). And FYI: I’ve checked some of the classic stuff on PeaksOnly. Unsurprisingly, the Evergreens just sound awesome on this mix check tool (that stuff wasn’t done using modern mix techniques: they had to balance the peak energies back in those days, and it shows). You can join me for that Podcast, if you’re a patron.

Date September 2, 2019
Author Chris
Comments 2 Comments

Chris

Biquad2

Channel7

Monitoring

PeaksOnly

handsewn bespoke digital audio

Kinds Of Things

The Last Year

Patreon Promo Club