Airwindows

Hit Record Meter (0.1.8)

TL;DW: Meter uses analysis of peak energy to rate songs for hit potential.

Here’s what’s new in version 0.1.8! Meter now focuses entirely on peaks (RMS is used for some internal calculations and the size of the dots in the Zero Cross section).

Each section is labeled: Peak Loudness, Slew Brightness, Zero Cross Bass. They all show red, blue and green dots. It’s always the same data, just arranged differently, like this…

Peak Loudness shows the dots on a dB meter (labelling the horizonal lines as -6, -12, -18, -24, and -30 dB). This is the same as a normal RMS meter, except it’s only showing peaks: if they are not showing up at the bottom of the meter, the RMS loudness is too loud to let them go down there.

Slew Brightness arranges the dots by slew factor, so brighter ones will be higher (as a rule).

Zero Cross tracks how long the audio could go before crossing zero, so this is not only presence of low bass, it’s also about whether there are higher frequencies to interfere with that bass. It’s also labelled now, at 200 hz, 40 hz and 20 hz (which is the bottom of the meter). Again, it tracks not just whether bass exists, but whether it’s allowed to dominate. If you notice, there are lines higher than 200hz around where the ‘Zero Cross’ label is: those lines are 2k and 20k, and most audio shouldn’t even get near there for zero crossings. Refer to music you like as a reference for how this ought to look.

There’s a line of text now that tells you about three things: the original Loudness measurement (which isn’t RMS, it’s the raw density of how many peaks are present), a new measurement for novelty (how much the pattern of peaks changes, making a different sound), and a measurement of how many bright, loud, and dark peaks happened. Dark peaks aren’t always bass, they’re just peaks where the slew isn’t high enough, just as bright peaks are all slew and treble. Meter now keeps track of this to tell you if you’re over-bright or over-dark.

And there’s a rating, like there was before. But now it’s not ‘peak loudness’. Now it’s novelty MINUS peak loudness and MINUS how off-balance the bright/dark peaks are. The idea was to track down roughly how striking the sounds were, even though Meter doesn’t know what a note is or understand music per se. Turns out, this new Meter is very good at singling out breakthrough songs that broke a big act (for instance, its favorite Led Zeppelin track is Good Times Bad Times) and career-making records like Sergeant Pepper. It likes punchy, dynamic music like the B-52s and the Beastie Boys and Chic. Its favorite Aerosmith track seems to be Walk This Way, and it’s sorted the Yes tracks I’ve so far recorded, into a list that is almost exactly sorted by record sales.

If you think that’s interesting and want to mix stuff to make Airwindows Meter happy, the results you get will probably sound good once you’re done. I can’t make it give you Top Ten hits, that stuff was back when we had a record industry. But it can help you get striking and exciting sounds. You can also use it to match other music you know: study the meters and make your music match what you see on your target music and that should help. But to pursue hits as Meter understands them, allow for a bunch of headroom and then use up ALL that space with peak energy. It likes things a little dry, not loads of reverb, and it likes it when the arrangement leaves spaces: if possible, space like the song is breathing in tempo with the desired music. Definitely pay attention to whether bright/dark is out of balance, but you can either go for the hi-fi sound of bright/dark peaks, or you can just try to make all the sounds peak out as loudly as possible, which means mixing everything to be loud and sonorous. Both work.

There’s more tools coming for working with all this, but this is a good update for being able to keep track of what you’re doing with all those plugins. If Meter breaks or fails to work for you, I’ll try to get help as to fixing it: I’m out over my skis working with JUCE but with the help of the Pamplejuce framework, I can try to provide GUI plugins. Have fun!

Date October 6, 2024
Author Chris
Comments 7 Comments

ToTape8

TL;DW: ToTape8 is Airwindows tape emulation with gain staging, bias and optimized Dubly!

ToTape8 in Airwindows Consolidated under ‘Tape’ (CLAP, AU, VST3, LV2)
ToTape8.zip(569k) standalone(AU, VST2)

Sometimes, things move kinda fast.

ToTape8 comes right on the heels of ToTape7, rather than a couple of years later. Why? Necessity. So much was right and new with ToTape7, but it was letting itself down. It wasn’t clear how to set levels on it as there was only ‘TapeDrv’ in the middle of many other controls. There was lots of control over the Dubly, but it’d never made its way into a ToTape version before and quickly showed a gritty, fierce character that actually limited the amount of good Dubly could even do with the tone. Yet it took up four controls, getting in the way of an output level control while still allowing use with Consolidated and the VCV Rack version.

Clearly, the answer was to immediately get COVID! But in recuperating from that, there was time to revisit ToTape and Dubly, and the new Dubly is out, and here’s the next generation of ToTape. Already. And probably about two years worth of ‘better’, if you’re wondering, because that was just waiting to be unleashed: all the other stuff was ready to support it.

It’s now obvious how to gain stage ToTape, with an input and an output control… but it’s more sophisticated than it looks, because ToTape8 has a safety clipper built in (same as ToTape7). This is ClipOnly2, running after ToTape so no matter what you do the output is safe for a final mixdown… but since there’s an output control now, the output control goes after the tape emulation and before the final clip. So if you want to pull it back to ‘tape overdrive only’, you can set the output to about 0.4 or so (depends on your other settings), shy of the ClipOnly2 clip. Or, if you want a much cleaner tape effect with more powerful peaks, leave it as default or even turn it up: the Input control can have you hit the ‘tape’ less hard, and since idealized tape needn’t have noise you can gain stage it any way you want (there can be a kind of noise in a future variation that’s more lo-fi, this is more for being able to put on tracks and mix buss without wrecking the tone)

With no tape noise, what does the new Dubly do? Amazing things. Now that it does a kind of compression (not a clone of hardware, but a simpler, optimized compression to clean up the grit) that’s tied with the Dubly frequency, the formerly crunchy Dubly has been reduced to two controls, but they’re amazing and way more useful than the original four. Tilt simply adjusts the Dubly amounts against each other. Maybe 0.5 is not, in fact, the perfect setting. Try other settings! Anything close to the center will be musical and useful, cranking it to 1.0 will give you the ‘Dubly encode trick’ used in studios, and turning it down to 0.0 gives you only decode, at the most extreme amount possible. So, it’s named Tilt because it works like a tilt EQ, but it comes from the existing Dubly. Then, the Shape control varies the cutoffs of the Dubly encode and decode, AND the response of the compressors, against each other. What this does might not be obvious, but once you try it, there’s nothing more obvious. Tweaking this adjusts the midrange hype. You can have the mids be more searing and compressed and lively, or you can dial them back until they’re dark and vibey. You get a funky expanded dynamic tone you’ll immediately recognize from certain Seventies records. It’s at your fingertips with the Shape control.

Does that mean setting it to neutral does nothing? Nope, this is the best part. Just by being there, Dubly does a special thing with the ToTape overdrive/saturation. It cancels harmonics, and this is why real tape machines are often thought of as near-magical ‘clean level compressors’. ToTape8 pulls this off pretty well, even though it’s minimal and clean enough to use in real audio work (not just sound design). Note that this is not ‘aliasing control’ though it’ll sound like it is, compared to the gritty ToTape7 (vibey though it is). ToTape8 does use filtering in both the Dubly encode and decode, but not in an ‘elaborate brickwall filter because it thinks it is oversampling’ way, in a way that’s designed for better tone and ability to run at zero latency. What’s happening is, the nature of Dubly means anything producing harmonics tends to also have the harmonics taken back out again, because they’re treated as additional noise. It’s that simple, and still works in minimal form in spite of not cloning any hardware box, and that’s why ToTape8 gets the sound it does.

That’s the new stuff, Flutter and Bias and Head Bump and Head Bump Frequency still work the way they do in ToTape7. I think they’ll be even easier and more pleasing to work with now that the Dubly stuff has been brought up to par. Let me know if this plugin sounds better than ToTape7. It ought to :)

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.

Date September 29, 2024
Author Chris
Comments 11 Comments

Dubly3

TL;DW: Dubly3 refines and transforms the Dubly sound.

Dubly3 in Airwindows Consolidated under ‘Effects’ (CLAP, AU, VST3, LV2)
Dubly3.zip(506k) standalone(AU, VST2)

So here’s where the Dubly stuff suddenly gets more useful for a lot of people.

This whole time, Dubly’s been using a transfer function called uLaw encoding to take the place of ‘compression’, because it gets to roughly the same 2:1 ratio that the famous noise reduction system liked to use. And this is all well and good… but transfer function modifications without time constants aren’t compression. They’re distortion. And so all the versions of Dubly have used brightened but distorted audio to encode and decode with. And this is fine for a number of reasons (one of them being, Dubly is a Spinal Tap joke and not an attempt to pirate still very defended IP) up until ToTape7 came out, and people started trying to use it for all manner of things.

Dubly and Dubly2 are the right basic sort of thing, but gritty and alias-y, and I vowed I’d work on it in order to develop a ToTape8 that performed better, ideally with a heck of a lot less aliasing but still zero latency and with the immediacy Airwindows ToTape has been known for. And so, towards that goal, Dubly3 is here.

What’s changed? A bunch of things. Notably, Dubly3 uses real compression: but not in any sense a clone of other gear. Instead, it’s a new trick I might look into further: it’s applying a compression based not on the input signal, but the uLaw version of the input signal, so the ‘DNA’ of original Dubly is very much still in there. But the time constants totally alter the texture of how Dubly3 behaves. It acts a lot more like a real compressor, as in a sense it is one, just one that’s following an unusual form of signal (much like having the compression try to follow a tube-drive style control signal).

That applies to both encode and decode. On top of that, Dubly3 does adopt the classic trick of clipping the compressed highs slightly, which also helps it sound authentic. But beyond that, Dubly3 reinvents the controls in a way that I think will be very helpful, and in a way that lets the upcoming ToTape8 add some asked-for features.

Instead of direct control over amounts of encode, decode, and their crossover frequencies, Dubly3 boils it down to input and output gains (labeled as such, not ‘tape drive’) and tilt and shape controls, and that’s how you dial it in. What’s tilt? At 0.5 tilt is ‘neutral’, but as you boost it, you’re adding encode and cutting decode. The similarity to a tilt EQ is obvious. More tilt means more bright, and if you cut it below 0.5 you’re leaving decode fully engaged but reducing encode, causing it still to be the Dubly sound, but darker.

Shape is even more interesting. Where Tilt balances the encode and decode levels against each other, Shape balances the crossover frequencies against each other… and lets you push them to unreasonable extremes, in case it gives you usefully weird sounds to play with. For real sounds you’ll be keeping it a lot nearer 0.5 where it defaults to. What happens is this: if you increase Shape, you’re encoding at lower frequencies, but decoding at higher. That means the brightness factor stays roughly the same, but the mids get more intense, more compressed, more lively. You’ll hear it in the chug of guitars and in the energy of percussion, but it’s not purely a treble hype: it’s reshaping the whole midrange where Dubly crosses it over to treble. Then, when you cut Shape, you’re encoding at higher frequencies but decoding more in lower frequencies… and since decoding is more subtractive, and since Dubly3 is a compressy thing, you’re darkening and expanding the mids instead.

Using these controls, you can use Dubly3 to start dialing in a number of classic record sounds, particular albums where perhaps the hardware noise reduction of the day had drifted and produced a distinctive sound. And then, once ToTape8 is done, you can apply it to the additional controls of ToTape, adding refinements in bias and using the classic ToTape algorithm to make a plugin you can hit like it was a real tape machine.

Which real tape machine? ANY real tape machine, pretty much. If you can identify the sound (things like where the head bump sits, how overbiased or underbiased the sound is, whether it’s a brighter or darker sound, more saturated and hyped in the mids or cleaner and more dark and studio-y… at that point it ought to be possible to decode the classic sounds and either mimic them, or learn from them and use the plugin to make the choices that work for YOUR mix and your sound.

It starts with Dubly3, and once I’ve got ToTape8 done, you’ll have everything ToTape7 has, except with this Dubly instead of the previous, and thanks to the use of two fewer tone controls, ToTape8 will also have an Input control at the top and an Output control at the bottom, to make the gain staging simpler for people to deal with. And it will still fit in Airwindows Consolidated. But for now, play with Dubly3 and see if that works better for ya than Dubly2 did. And Dubly2 is still supported and available because it’s an entirely different sound and you should still have it in case you can use it.

Next, a ToTape8, but in weeks rather than another couple years. And then… more work on ConsoleX :)

Date September 22, 2024
Author Chris
Comments 1 Comment

Console9

TL;DW: Console9 is just the summing from ConsoleX.

Console9 in Airwindows Consolidated under ‘Consoles’ (CLAP, AU, VST3, LV2)
Console9.zip(993k) standalone(AU, VST2)

Here we have the first part of ConsoleX that can come out and be put into use, just as it will be used in ConsoleX.

You see, the upcoming Console system is the big one with all the EQs and multiple dynamics paths and StoneFire built in, and it’s got so much going on it requires the development of an elaborate GUI (for which a lot of work has already been done). But the guts of the system remains Airwindows Console and that gets its own development. Console is the system by which all the sound sources get pre-distorted and then un-distorted on the buss after summing, for a sound that (in theory) doesn’t touch the tones of any individual mix element, but alters the spaces between the sounds.

So when using Console (or any version like Console9 that’s strict about this) you should expect to hear exactly the same sounds you originally had, but they’ll sit differently. They’ll make up a more convincing space, in a subtle way that shouldn’t mess with your sounds, that will just make it easier to make those sounds gel into a mix that feels good.

The algorithms for Console9 (as used in ConsoleX) are new, but they came about through work I did revising Console6. It turns out there’s a resource called ‘Herbie’ that helps you re-hack algorithms to work better in floating point. That’s immediately of interest to me, as I’m fascinated by the ways our simple digital math erodes and harms the tone of our sounds… I’ve already done a lot of work on word length and dithering, and I already advocate double precision processing over the use of simple floats. But it turns out algorithms that don’t play right by Herbie can have special (bad) qualities, and the original Console6 plugins were guilty of this.

What would happen is there’d be a big spike of inaccuracy as the calculations swept by zero. And that’s as may be, but we’re dealing with audio: especially with low frequencies, the sound is constantly sweeping through zero! It’s digital math reinventing how to make crossover distortion as found in class AB amplifiers. Herbie exists to re-design such algorithms so they sweep through zero (or wherever else there’s an issue) perfectly. It’ll also estimate how much extra CPU it’ll take: sometimes it’ll do a silly approximation and say ‘oh hey, if you just replace all the outputs with 0 you still have X much accuracy but the code runs much quicker!’

There were optimizations for Console6, and the version in Consolidated (and in the downloadable plugin collections) is the new version. The download from when Console6 first came out, remains the same for historical purposes and so you can get a hold of exactly how it was when it came out (this is true for pretty much all Airwindows plugins: use the collections or Consolidated to have the recentest versions).

But it turned out there was also a set of distort/undistort curves available from using the same type of math as in Console6, the Herbie-optimized sort, except scaling things by the golden ratio (of course it had to be the golden ratio, that thing pops up all over the place) and this produced a new sound for Console. It’s got more headroom than any previous Console system, while also having some of the raw intensity of Console6 (which, even fixed, might be a little too raucous and garagey for most purposes). It’s got a real good sense of transparency but it strongly brings that Console vibe to the spaces between the notes. And it’s the summing buss that is going into ConsoleX.

The reason I say it’s also the part that will be used IN ConsoleX, is because I’m adopting some Chris Lord-Alge-isms in ConsoleX. My take on how sends and verbs and things should be handled in ConsoleX is to use them kind of like Console8 uses submixes: you’ll be sending stuff from channels to other channels that have the verbs on them.

The difference is, when you do that you put Console9Buss at the top of the sent-to channel… put in your verb or whatever, all wet… and then run the result through ANOTHER instance of ConsoleXChannel, and sum that. So, the idea is ‘decode and then recode, except you get literally all the processing from any ConsoleXChannel, all over again, on the verb or delay returns’. And then you adjust those the same way you’d adjust any other channel: they just come in on even more channels, and all the same features apply. And so, Console9Buss is your ‘decode’ for making further verb stuff happen, because you don’t need to run the full ConsoleXBuss to do that.

There’s still loads of work to get ConsoleX done, plus I’m having another crack at ToTape to address the eager aliasing that came out of Dubly2. Unlike that weird beast, Console9 is very clean and simple so you can oversample it to the moon if you wanted to: I will not be doing that, I’ll be running it at 96k and it will be fine. There is a reason people got an ‘analog immediacy’ hit off ToTape7: aliasing suppression is a tradeoff between artifact minimization and overprocessing tone-flattening, and in ToTape you’re hearing one extreme. I’m still going to see whether I can have best of both worlds but I’m not taking away the ToTape7 people enjoyed. And when ConsoleX comes out, there’ll be a dedicated ‘only the tone controls’ plugin (for use pre-EQing stuff to a channel instance of ToTape, which is a black art of its own) and so you’d be able to run the EQs and then oversample the summing buss to the moon if that pleases you, using Console9 to stand in for the built-in versions that are part of ConsoleX.

For now, I will continue to get over my recent bout of COVID (I’m testing negative, have just lost a lot of energy so be patient), will continue working through the backlog of plugins that can come out, and I’ll talk to ya later :)

Date September 15, 2024
Author Chris
Comments 8 Comments

Chris

Hit Record Meter (0.1.8)

ToTape8

Dubly3

Console9

handsewn bespoke digital audio

Kinds Of Things

The Last Year

Patreon Promo Club