funasr pages

游雁

2023-11-16 4ace5a95b052d338947fc88809a440ccd55cf6b4

 setup.py

@@ -4,7 +4,6 @@

import os

from distutils.version import LooseVersion
from setuptools import find_packages
from setuptools import setup

@@ -12,71 +11,54 @@
requirements = {
    "install": [
        "setuptools>=38.5.1",
        # "configargparse>=1.2.1",
        "typeguard>=2.7.0",
        "humanfriendly",
        "scipy>=1.4.1",
        # "filelock",
        "librosa>=0.8.0",
        "jamo==0.4.1",  # For kss
        "librosa",
        "jamo",  # For kss
        "PyYAML>=5.1.2",
        "soundfile>=0.10.2",
        "h5py>=2.10.0",
        "soundfile>=0.12.1",
        "h5py>=3.1.0",
        "kaldiio>=2.17.0",
        "torch_complex",
        "nltk>=3.4.5",
        # ASR
        "sentencepiece",
        # "ctc-segmentation<1.8,>=1.6.6",
        "jieba",
        "rotary_embedding_torch",
        "ffmpeg",
        # TTS
        # "pyworld>=0.2.10",
        "pypinyin<=0.44.0",
        "pypinyin>=0.44.0",
        "espnet_tts_frontend",
        # ENH
        # "ci_sdr",
        "pytorch_wpe",
        "editdistance==0.5.2",
        "tensorboard==1.15",
        "editdistance>=0.5.2",
        "tensorboard",
        "g2p",
        "nara_wpe",
        # PAI
        "oss2",
        "kaldi-native-fbank",
        "edit-distance",
        "textgrid",
        "protobuf",
        "tqdm",
        "hdbscan",
        "umap",
    ],
    # train: The modules invoked when training only.
    "train": [
        # "pillow>=6.1.0",
        "editdistance==0.5.2",
        "editdistance",
        "wandb",
    ],
    # recipe: The modules actually are not invoked in the main module of funasr,
    #         but are invoked for the python scripts in each recipe
    "recipe": [
        "espnet_model_zoo",
        # "gdown",
        # "resampy",
        # "pysptk>=0.1.17",
        # "morfessor",  # for zeroth-korean
        # "youtube_dl",  # for laborotv
        # "nnmnkwii",
        # "museval>=0.2.1",
        # "pystoi>=0.2.2",
        # "mir-eval>=0.6",
        # "fastdtw",
        # "nara_wpe>=0.0.5",
        # "sacrebleu>=1.5.1",
    ],
    # all: The modules should be optionally installled due to some reason.
    #      Please consider moving them to "install" occasionally
    # NOTE(kamo): The modules in "train" and "recipe" are appended into "all"
    "all": [
        # NOTE(kamo): Append modules requiring specific pytorch version or torch>1.3.0
        "torch_optimizer",
        "fairscale",
        "transformers",
        # "gtn==0.0.0",
    ],
    "setup": [
        "numpy<=1.21.3",
        "numpy",
        "pytest-runner",
    ],
    "test": [
@@ -93,17 +75,18 @@
        "black",
    ],
    "doc": [
        "Jinja2<3.1",
        "Sphinx==2.1.2",
        "Jinja2",
        "Sphinx",
        "sphinx-rtd-theme>=0.2.4",
        "sphinx-argparse>=0.2.5",
        "commonmark==0.8.1",
        "commonmark",
        "recommonmark>=0.4.0",
        "nbsphinx>=0.4.2",
        "sphinx-markdown-tables>=0.0.12",
        "configargparse>=1.2.1"
    ],
}
requirements["all"].extend(requirements["train"] + requirements["recipe"])
requirements["all"].extend(requirements["train"])
requirements["test"].extend(requirements["train"])

install_requires = requirements["install"]
@@ -121,7 +104,7 @@
    name="funasr",
    version=version,
    url="https://github.com/alibaba-damo-academy/FunASR.git",
    author="Speech Lab, Alibaba Group, China",
    author="Speech Lab of DAMO Academy, Alibaba Group",
    author_email="funasr@list.alibaba-inc.com",
    description="FunASR: A Fundamental End-to-End Speech Recognition Toolkit",
    long_description=open(os.path.join(dirname, "README.md"), encoding="utf-8").read(),