TurboRust Audio Next 01 — offline candidate, 2026-11-04

Music and SFX: original Codex-authored compositions, arrangements and mathematical
synthesis for this project. No reference-game recordings or melodies imported.
Speech: original scripts synthesized locally with hexgrad/Kokoro-82M v1.0 and
synthetic stock voices am_michael, am_puck, bm_george, af_heart. The pinned model
card labels weights Apache-2.0; the official code license is included. Input URLs,
revisions and hashes are in provenance/kokoro_inputs.json. No cloned performer.
Source: https://huggingface.co/hexgrad/Kokoro-82M
Code: https://github.com/hexgrad/kokoro

Offline processing tools and models are not redistributed in the game asset bank.
Environment locks identify the tools used, not additional runtime dependencies.
No ACE-Step output is included: setup stopped before inference. Prepared prompts
are included for an optional future comparison. Model licenses do not themselves
establish that every generated output passes an originality/listening review.
