| New file |
| | |
| | | # 安装 |
| | | FunASR的安装十分便捷,下面将给出详细的安装步骤: |
| | | |
| | | - 安装Conda并创建虚拟环境 |
| | | ``` sh |
| | | wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh |
| | | sh Miniconda3-latest-Linux-x86_64.sh |
| | | conda create -n funasr python=3.7 |
| | | conda activate funasr |
| | | ``` |
| | | |
| | | - 安装Pytorch (版本 >= 1.7.0): |
| | | |
| | | ```sh |
| | | pip install torch torchvision torchaudio |
| | | ``` |
| | | |
| | | 关于更多的版本, 请参照 [https://pytorch.org/get-started/locally](https://pytorch.org/get-started/locally) |
| | | |
| | | - 安装 ModelScope |
| | | |
| | | 对于国内用户,可以通过配置下述镜像源来加快下载速度 |
| | | ```sh |
| | | pip config set global.index-url https://mirror.sjtu.edu.cn/pypi/web/simple |
| | | ``` |
| | | |
| | | 安装或更新ModelScope |
| | | ``` sh |
| | | pip install "modelscope[audio]" --upgrade -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html |
| | | ``` |
| | | |
| | | - 下载FunASR仓库,并安装剩余所需依赖 |
| | | ``` sh |
| | | git clone https://github.com/alibaba/FunASR.git && cd FunASR |
| | | pip install --editable ./ |
| | | ``` |