Ableton/Live/Instrument/Drum/Forevertron/Log/26/11/08-Sun

• 2026-11-08 Sun
  • AI/Response from Anthropic/Model/Claude/5/5/Opus

    • The problem

      • Most pads in the Forevertron Ableton/Instrument/Drum Rack stopped when the note was released. The fix is to put each pad's Simpler on Trigger, as Ableton/Instrument/Drum Rack/Q/How can I make all the drums in my drum rack be the kind where I press the note and they keep playing? explains. Live 11 has no control that changes 76 pads at once, so the preset file was edited directly instead.

    • Where the samples are

      • All 76 samples are recordings from Ableton/Note, synced to the Mac by Ableton's cloud: ~/Library/Application Support/Ableton/Cloud Manager/d523cca8-7229-4f3f-b488-dfaf42d5c379/Forevertron/Samples/Recordings/
      • They're named 29_06_24-R1.wav to R5 and 30_06_24-R1.wav to R79, recorded June 29 and 30, 2024, at 48 kHz. Each has a .asd analysis file beside it. None are missing.
      • ~/Library is hidden, so Live's browser doesn't show this folder unless it's added under Places.

    • Three presets

      • All three are in ~/Music/Ableton/User Library/Presets/Instruments/Drum Rack/.
      • Forevertron.adg — the original rack, saved from Live 11.3 on November 8 at 14:51. Each pad is a Simpler in One-Shot mode. Three pads are on Trigger (29_06_24-R1, 29_06_24-R2 and 30_06_24-R4), and the other 73 are on Gate.
      • Forevertron Trigger.adg — the same rack with all 76 pads on Trigger. Nothing else differs.
      • Forevertron Trimmed.adg — Forevertron Trigger, plus new start, end and fade-out points on 39 of the 40 pads that still played from the very start of the file. The 36 pads trimmed by hand are unchanged.

    • How the preset file stores Trigger and Gate

      • An .adg file is gzipped XML. Each pad's Simpler is an OriginalSimpler element.
      • Trigger and Gate are OneShotEnvelope/SustainMode: 0 is Trigger, 1 is Gate. This was confirmed by saving the rack with 29_06_24-R2 on Trigger and 29_06_24-R3 on Gate and comparing the two pads.
      • The start and end points are SampleStart and SampleEnd, counted in samples. Fade Out is OneShotEnvelope/FadeOutTime, in milliseconds, and Simpler caps it at 2000.

    • How the 39 pads were trimmed

      • Start: about 3 ms before the drum hit's attack, moved back to the nearest zero crossing.
      • Choosing a hit: when a recording has several hits, only those within 10 dB of the loudest count. Of those, the one that rings longest before the next hit or the background noise wins.
      • End: 1 s after the hit fades into the background noise, 20 ms before the next hit, or the end of the file, whichever comes first.
      • Fade Out: half the time between start and end, up to 2000 ms.
      • Compared with the hand-trimmed pads: on 35 hand-trimmed pads, the detector picked the same hit 29 times. It started 2–6 ms earlier than the hand-set start on most of them, and about 20 ms later on 29_06_24-R1, 29_06_24-R2, 30_06_24-R3 and 30_06_24-R37, where the hand-set start leaves extra room before a soft attack. The hand-set endings usually run longer than the detected ones.
      • Skipped: 29_06_24-R5 is background noise all the way through, peaking only 4 dB above it, so it has no hit to start on.
      • Worth a listen first:
        • Several hits of similar strength, so the choice is a guess: 30_06_24-R9, R19, R20, R23, R34, R53, R57, R76 and R79.
        • Quiet recordings, 22–25 dB above the background noise: 30_06_24-R55, R71, R73 and R77.