New tts/generate_sfx.py mirrors generate.py (MOSS-TTS) but for the
MOSS-SoundEffect checkpoint: text-only ambience/SFX generation, no
reference audio or voice cloning. Needed PYTORCH_ALLOC_CONF=
expandable_segments:True baked in - this checkpoints weights alone use
~22.3GB of this boxs ~23.56GB usable VRAM, leaving generate() itself no
room without it (confirmed via OOM otherwise).
webui/app.py gets a new standalone "5. Sound effects" section in the
Pipeline Studio tab (separate from the AuK/MOSS-TTS voice flow - this is
a different generation primitive with no reference/history state).
Verified end-to-end: three real test clips (door creak, market ambience,
car pass) generated cleanly, approved by Steve as "definitely passable,"
then wired into the GUI and confirmed working through app.py itself.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01SN15Vxn4svKuhYxFsVtMJf
--instruction passes through to MOSS-TTS's build_user_message (confirmed
to work for emotion in earlier testing) so callers can steer per-line
delivery. --seed enables reproducible generation, currently the only
way to get determinism out of this pipeline since MOSS-TTS/AuK don't
use a fixed-seed workflow the way Qwen3 did.
Needed for a downstream per-line episode-generation pipeline (character
cards + emotion-tagged script lines).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01Y1oQ4h4WE4R9KRvMQLr5Ey
Gradio app that orchestrates the three existing isolated venvs via
subprocess (their dependencies conflict too much to merge into one
environment - MOSS-TTS needs transformers 5.0.0, AuK needs 4.57.6).
Lets you strip a voice, iteratively edit it (emotion/pitch/speed/custom
instruction) with preview and undo/reset after each step, then generate
new speech in the resulting voice - addressing the trial-and-error
workflow the pitch-overshoot investigation needed.
Also extracts tts/generate.py as a reusable, general-purpose MOSS-TTS
cloning CLI (text/reference/language/output args) rather than one-off
personal scripts.
Verified all three stages end to end via direct function calls before
wiring up the UI.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01Y1oQ4h4WE4R9KRvMQLr5Ey
steve-adminandClaude Sonnet 5