pyproject.toml

[project]
name = "asr-lab"
version = "1.0.0"
description = "Benchmarking platform for automatic speech recognition models using controlled audio degradation, multiple evaluation metrics, and comparative analysis across languages and model architectures."
readme = "README.md"
requires-python = ">=3.12"
dependencies = [
    "accelerate>=1.10.1",
    "click",
    "datasets>=4.2.0",
    "evaluate>=0.4.6",
    "jiwer>=4.0.0",
    "joblib",
    "librosa",
    "matplotlib>=3.10.7",
    "pandas",
    "pedalboard",
    "plotly>=5.24.0",
    "pyaudio>=0.2.14",
    "pyloudnorm",
    "seaborn>=0.13.2",
    "jinja2",
    "torch",
    "transformers>=4.44.0",
    "vosk>=0.3.45",
    "torchaudio>=2.8.0",
    "sentencepiece>=0.2.1",
    "protobuf>=5.29.5,<5.30",
    "nemo_toolkit[asr]>=2.0.0",
    "tqdm>=4.66.0",
    "ruff>=0.4.0",
    "hf-xet>=1.2.0",
    "pydantic>=2.12.4",
    "demucs>=4.0.1",
    "diffq>=0.2.4",
    "funasr>=1.1.0",
    "modelscope>=1.9.5",
]

[project.optional-dependencies]
enhancement = [
    "deepfilternet>=0.5.6",
]

[build-system]
requires = ["setuptools", "wheel", "Cython"]

[tool.uv]
override-dependencies = [
    "triton; sys_platform != 'win32'",
]