fastapi
librosa
soundfile
uvicorn
websockets
huggingface-hub>=0.25.0
faster-whisper>=1.2.0
torch>=2.0.0
torchaudio>=2.0.0
tqdm
tiktoken

[cpu]
torch>=2.0.0
torchaudio>=2.0.0

[cu129]
torch>=2.0.0
torchaudio>=2.0.0

[cu129:platform_machine == "x86_64" and (sys_platform == "linux" or sys_platform == "linux2")]
triton>=2.0.0

[diarization-diart]
diart
torch<2.9.0
torchaudio<2.9.0
torchvision<0.24.0

[diarization-sortformer]

[diarization-sortformer:python_version >= "3.10" and python_version < "3.13"]
nemo-toolkit[asr]>2.4

[listen]
sounddevice>=0.4.6

[mlx-whisper]

[mlx-whisper:sys_platform == "darwin" and platform_machine == "arm64"]
mlx>=0.11.0
mlx-whisper>=0.4.0

[sentence_tokenizer]
mosestokenizer
wtpsplit

[test]
pytest>=7.0
pytest-asyncio>=0.21
datasets>=2.14
librosa

[translation]
nllw

[voxtral-hf]
mistral-common[audio]
accelerate>=0.12

[voxtral-hf:python_version >= "3.10"]
transformers>=5.2.0

[voxtral-mlx]
mistral-common[audio]

[voxtral-mlx:sys_platform == "darwin" and platform_machine == "arm64"]
mlx>=0.11.0
mlx-whisper>=0.4.0
