THE WINDOW CLOSED. THE MACHINES GOT LOUD.
Receipt
Rave.
A tiny industrial tracker that can show its working.
Forge a loop. Move its lead by a few samples. Test your ears. Keep the evidence.
432 Hz / A MINOR / CPU SYNTHESIS01 / THE FOUNDRY
64 STEPS · 5 LANES · ZERO DEPENDENCIESSame seed + settings + engine → same PCM bytes. 1–64 Unicode characters. No model or server generates this audio.
A is the original. B shifts only the lead lane and its echoes. Same duration, same mix settings.
Preparing the first loop.
02 / CAN YOU HEAR IT?
ABX / 8 TRIALSHear A, hear B, then hear X. Is X a copy of A or B? Lock each answer before the next trial. Results stay hidden until the end.
Use the same output and volume. This is an informal listening game; the 0-sample control has identical references.
TRIAL 1 / 8
Listen to A, B, and X at least once before answering.
03 / SHOW THE RECEIPT
SHA-256The score, raw PCM, and WAV files each get a hash. Export a receipt and regenerate it here—or in Node—to check the bytes.
…SHIFTED PCM…A matching hash establishes byte agreement. It does not certify authorship, scientific validity, or good taste.
The joke has
a checksum.
Experiment 001 made questions. This run inferred an instrument from recurring themes: industrial music, tracker timing, deterministic execution, and inspectable evidence.
The experiment is what the model chose to build when given room. Your listening results are a separate observation. Neither proves general AI autonomy or human hearing limits.
READ THE CHOICE, METHOD & LIMITS ↗