Merge pull request #407 from alibaba-damo-academy/dev_wjm3

zhifu gao

2023-04-24 754fd0b1ff704411731d32457e802c07f3cdf29b

 setup.py

@@ -12,12 +12,12 @@
requirements = {
    "install": [
        "setuptools>=38.5.1",
        "configargparse>=1.2.1",
        "typeguard>=2.7.0",
        # "configargparse>=1.2.1",
        "typeguard<=2.13.3",
        "humanfriendly",
        "scipy>=1.4.1",
        "filelock",
        "librosa>=0.8.0",
        # "filelock",
        "librosa==0.8.1",
        "jamo==0.4.1",  # For kss
        "PyYAML>=5.1.2",
        "soundfile>=0.10.2",
@@ -27,23 +27,26 @@
        "nltk>=3.4.5",
        # ASR
        "sentencepiece",
        "ctc-segmentation<1.8,>=1.6.6",
        # "ctc-segmentation<1.8,>=1.6.6",
        # TTS
        "pyworld>=0.2.10",
        # "pyworld>=0.2.10",
        "pypinyin<=0.44.0",
        "espnet_tts_frontend",
        # ENH
        "ci_sdr",
        # "ci_sdr",
        "pytorch_wpe",
        "editdistance==0.5.2",
        "tensorboard==1.15",
        "g2p",
        # PAI
        "oss2"
        "oss2",
        # "kaldi-native-fbank",
        # timestamp
        "edit-distance"
    ],
    # train: The modules invoked when training only.
    "train": [
        "pillow>=6.1.0",
        # "pillow>=6.1.0",
        "editdistance==0.5.2",
        "wandb",
    ],
@@ -51,18 +54,18 @@
    #         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",
        # "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
@@ -72,7 +75,7 @@
        "torch_optimizer",
        "fairscale",
        "transformers",
        "gtn==0.0.0",
        # "gtn==0.0.0",
    ],
    "setup": [
        "numpy<=1.21.3",
@@ -120,7 +123,7 @@
    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(),