revistasobreorugas/faster-qwen3-tts-demo

add demo continuation comparison

Match text-feed defaults to upstream modes

Add demo text-feed toggle for non_streaming_mode

fix(demo): don't switch uploads to Simple mode

fix(demo): default clone toggle to Advanced (ICL) mode

fix(demo): prevent stale preset from overriding browser recordings in ICL mode

Add audio size limit (10 MB / ~1 min) and restore text length guard

Add 1000-char text limit to prevent CUDA static cache overflow

Fix stale model ref race, add ACTIVE_MODELS env var, hide unavailable modes

Fix permission error for preset assets and PyTorch cache dir in HF Spaces

Make model cache size configurable via MODEL_CACHE_SIZE env var (default 2, Space uses 5)

Cache all loaded models (LRU, max 5) to avoid re-capture on model switch

Serialize generation with asyncio.Lock and show queue position to users

Center header and content together as a unit in viewport

Center app content vertically in viewport

Download preset wav files from GitHub at startup instead of bundling them

Move preset buttons left of play, match ref-chip height

Inline preset buttons into the reference audio row

Default preloaded model to 1.7B-Base for voice cloning

Reorder demo UI, default to advanced cloning, auto-select first preset

Move preset buttons inline and rename labels

Add preset voice clone refs in demo

Preserve mic selection after recording

Fix ruff issues

fix(demo): default to Simple (x-vector) cloning mode

fix(demo): add torchaudio to requirements

fix(demo): unconditionally import and load nano-parakeet

fix(demo): cap generation at 30s (360 tokens at 12 Hz)

feat(demo): auto-transcribe reference audio with nano-parakeet

fix(demo): show recording preview immediately, resample in background

feat(demo): clone row UX improvements

feat(demo): arcade mode selector, auto-load 1.7B-CustomVoice default

docs(space): add model preloads

UI rework

chore: hf space dockerization

fix(demo): robust mic recording + 16k resample

feat(demo): custom voices, xvec toggle, recording, theme

ui(demo): drop init metric

ui(demo): simplify metrics, add advanced toggle

cleanup(demo): drop ws/worklet dead code; cache ref audio

fix(demo): show zero init on warm runs

fix(demo): reuse ScriptProcessor audio context

revert(demo): use scriptprocessor streaming path

fix(demo): ws meta+pcm framing

fix(demo): respect pcm length in ws frames

feat(demo): binary ws audio chunks

feat(demo): cache ref audio for ws

feat(demo): cache ref audio base64 for ws

fix(demo): keep audio graph alive with silent oscillator

fix(demo): warm audio on user gesture and set path