f023653c71
- add align_text_to_audio(): forced alignment without ASR, fast - add _StableTSModels cache; detect language via stable-ts when language=None - LRC stage: use existing text.txt, copy it if already LRC, fall back to ASR only - filter junk lyrics from tags and ASR hallucinations (promo phrases, tics) - mark empty ASR results with metadata.empty_text and skip them on resume - warn in GUI when LRC is run with transcription disabled - add mark_no_speech.py one-shot script for existing tracks
22 lines
552 B
Plaintext
22 lines
552 B
Plaintext
# GUI
|
|
PySide6>=6.6
|
|
|
|
# ASR
|
|
whisperx>=3.1
|
|
stable-ts>=2.16
|
|
onnx-asr>=0.4
|
|
|
|
# Recognition & lyrics
|
|
shazamlite>=0.1
|
|
mutagen>=1.47
|
|
requests>=2.31
|
|
|
|
# WhisperX pulls torch; for CUDA you must install the matching torch build, e.g.:
|
|
# pip install torch torchaudio --index-url https://download.pytorch.org/whl/cu121
|
|
# For cuDNN preload helper used in main.py:
|
|
# pip install nvidia-cudnn-cu12
|
|
|
|
# System dependencies (not pip-installable):
|
|
# ffmpeg — must be in PATH
|
|
# unblend — vocal separation CLI (Mel-Band RoFormer), must be in PATH
|