STEEL TOE STOIC — WORKOUT AUDIO CREDITS

Music: Technomania101 (fixed-tempo version)
Author: Fupi
Source: https://opengameart.org/content/technomania101-2000s-europop-electronic-dance-music
Original download: https://opengameart.org/sites/default/files/technomania101_notempochange.flac
License: CC0 1.0 Universal
License terms: https://creativecommons.org/publicdomain/zero/1.0/
The author's source page identifies this recording as CC0. CC0 permits
copying, adapting, distributing and commercial use without permission.
No endorsement by the musician is implied.

Our derivative: a 32-second excerpt beginning at source second 48;
downmixed to mono at 22,050 Hz, peak-normalized, and wrap-crossfaded over
0.2 seconds. The full source recording and large models are not shipped.
Source FLAC SHA-256: fc370ce521f4d4593a118b74a79a7688c11b73a8d3170e35bb1ff01213c1dca6

Voice instructions: original concise workout direction text, synthesized
with preset voices am_michael (English) and em_alex (Spanish).
These are synthetic preset voices; no user voice was cloned.
Model: Kokoro-82M v1.0 by hexgrad, Apache License 2.0.
Model: https://huggingface.co/hexgrad/Kokoro-82M
Model license: https://huggingface.co/hexgrad/Kokoro-82M/blob/main/LICENSE
Model SHA-256: beb0d1848dee9a49da392cc3df26958d46cfa35d321edf434f52949153f0df3a
Voice vectors SHA-256: bca610b8308e8d99f32e6fe4197e7ec01679264efed0cac9140fe9c29f1fbf7d
ONNX distribution: https://github.com/thewh1teagle/kokoro-onnx/releases/tag/model-files-v1.1
Generator: kokoro-onnx 0.6.1 (MIT license).
Generator: https://github.com/thewh1teagle/kokoro-onnx
Generator license: https://github.com/thewh1teagle/kokoro-onnx/blob/main/LICENSE
Generated 2026-10-28. Output is mono 22,050 Hz, 16-bit PCM WAV.

Each movement cue is at most 8 seconds. Countdown numbers begin at
0, 1 and 2 seconds in a 3-second file. Full movement instructions remain
available in the workout interface. manifest.json includes text, duration,
paths, provenance and hashes. Regenerate with scripts/generate-workout-audio.py.

PROGRESSIVE TRAINING EXTENSION — 2026-10-28
Thirty exact exercise variants now have original concise English and Spanish
guidance using the same licensed Kokoro preset voices and signal-processing
pipeline described above. New speech files use the progressive filename prefix.
No user voice was cloned. The existing music, countdowns and completion cues
are preserved. Restriction-excluded movements deliberately share one generic
skip-and-rest recording per language; they never play the movement instruction.
Regenerate this additive vocabulary with scripts/generate-progressive-workout-audio.py.