nono.audio @noitsaudio

@Music7Ada Nice one to dig up. What did you make it in back then?

@PYRAMIDSKEME Granular on a music player is a great idea. How many grains a second are you running? Around 40 everything turns into a lovely smear, but the knobs start to feel like they do nothing.

@sethihq Love this. The ESP32 touch pins give you a number, not just on/off, so you might get velocity out of how much fruit your finger covers. I do something similar on iPhone glass, where a flatter finger plays louder.

@gavinski_s That matches how MIDI clock works: it only sends 24 ticks per beat, never a tempo, so the receiver has to estimate the BPM from the gaps. A sudden change throws that estimate off for a bit, which might be why resetting the transport fixed it.

@gavinski_s iPad-only makes sense for additive, there are a lot of partials to look at. Is editing them comfortable with a finger, or is it really a Pencil job?

@matrixsynth Always glad to see more talking-toy instruments. The 1978 chip never stored sound, just a tiny recipe for every 25 ms of speech, and rebuilt the voice each time. I think that's half the charm.

In C major that's the 7 in-key chords plus 5 off-key ones I picked by hand, like A♭maj7 and the backdoor B♭7. You can still recolour any chord after. nonoaudio.com/plugins/matcha…

The first Omnichord gave you a grid of chord buttons: pick a root, then major, minor or 7th. Matcha flips it. You pick the root and it picks the colour for your key. One m7 slid across all 12 roots is only in key 3 times.

Toussaint's paper, if you want the full list of rhythms: cgm.cs.mcgill.ca/~godfried/publ…

Confession: I don't run Bjorklund's algorithm at all. Mine is one line: step i is a hit if (i*k) mod n < k. Same shapes, sometimes starting on a different step. My first try shoved every hit to the end of its slot, and a test caught it the same evening.

The Euclid mode in Orbeat traces back to an algorithm Bjorklund came up with to space out timing pulses in a neutron accelerator. In 2005 Godfried Toussaint showed the same even spreading also spits out traditional rhythms.

Mine stores each word as ordinary audio, pre-rendered with Apple's text-to-speech. When it loads a word it works out a similar recipe itself, then plays it through the same kind of filter. A phone has room for the audio. The toy didn't. apps.apple.com/app/apple-stor…

SpellSynth's English word folder is 28 MB. The 1978 toy fit its whole built-in vocabulary in 32 KB of ROM, because the chip never stored sound, only a small recipe for every 25 ms of a word, and rebuilt the voice from it each time it spoke.

Acoustic took its slot and lives in the mids and highs instead, with a clicky struck attack on the kick and toms and brushy filtered noise on top. Lo-fi turned into Glitch the same day.

Orbeat started with three kits: Classic, 808 and Lo-fi. The 808 was out again about ten hours later. All its character was sub-bass, which a phone speaker can't really play, so it just sounded like Classic with effects on.

His rule: only the second 16th in each 8th moves late. 50% is straight, 66% a perfect triplet, and he says the good stuff lives between 50 and about 70. Only every other 16th moves, so the downbeats never shift.

Roger Linn's LM-1 sequenced at 48 ticks per beat. A swung 16th can only sit on a whole tick, so swing moves in jumps of about 4%: 50, 54, 58, 62, 66… GranSample uses his rule in 1% steps. Can anyone hear 54 vs 55?

Contour on the App Store: apps.apple.com/app/apple-stor…