Install the Python Package

You can select one of the following methods to install the Python package.

Method 1 (From pre-compiled wheels, CPU only)

Hint

This method supports the following platfroms:

• Linux (x64, aarch64, armv7l),

• macOS (x64, arm64)

• Windows (x64, x86)

Note that this method installs a CPU-only version of sherpa-onnx.

pip install sherpa-onnx sherpa-onnx-bin

To check you have installed sherpa-onnx successfully, please run

python3 -c "import sherpa_onnx; print(sherpa_onnx.__file__)"

which sherpa-onnx
sherpa-onnx --help

ls -lh $(dirname $(which sherpa-onnx))/sherpa-onnx*

Hint

You can find previous releases at https://k2-fsa.github.io/sherpa/onnx/cpu.html

For Chinese users and users who have no access to huggingface, please visit https://k2-fsa.github.io/sherpa/onnx/cpu-cn.html.

You can use:

pip install --verbose sherpa_onnx_bin sherpa_onnx_core sherpa_onnx --no-index -f https://k2-fsa.github.io/sherpa/onnx/cpu.html

or:

# For Chinese uers
pip install --verbose sherpa_onnx_bin sherpa_onnx_core sherpa_onnx --no-index -f https://k2-fsa.github.io/sherpa/onnx/cpu-cn.html

Hint

Make sure the logs contain information that it is downloading files from huggingface.co or hf-mirror.com.

The installation logs are shown below:

Looking in links: https://k2-fsa.github.io/sherpa/onnx/cpu.html
Collecting sherpa_onnx_bin
  Downloading https://huggingface.co/csukuangfj2/sherpa-onnx-wheels/resolve/main/cpu/1.12.23/sherpa_onnx_bin-1.12.23-py3-none-manylinux2014_x86_64.whl (21.6 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 21.6/21.6 MB 244.9 MB/s eta 0:00:00
Collecting sherpa_onnx_core
  Downloading https://huggingface.co/csukuangfj2/sherpa-onnx-wheels/resolve/main/cpu/1.12.23/sherpa_onnx_core-1.12.23-py3-none-manylinux2014_x86_64.whl (9.5 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 9.5/9.5 MB 207.6 MB/s eta 0:00:00
Collecting sherpa_onnx
  Downloading https://huggingface.co/csukuangfj2/sherpa-onnx-wheels/resolve/main/cpu/1.12.23/sherpa_onnx-1.12.23-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (4.1 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 4.1/4.1 MB 149.8 MB/s eta 0:00:00
Installing collected packages: sherpa_onnx_core, sherpa_onnx_bin, sherpa_onnx

Method 2 (From pre-compiled wheels, CPU + CUDA 11.8)

Note

This method installs a version of sherpa-onnx supporting both CUDA and CPU. You need to pass the argument provider=cuda to use NVIDIA GPU, which always uses GPU 0. Otherwise, it uses CPU by default.

Please use the environment variable CUDA_VISIBLE_DEVICES to control which GPU is mapped to GPU 0.

By default, provider is set to cpu.

Remeber to follow https://onnxruntime.ai/docs/execution-providers/CUDA-ExecutionProvider.html#requirements to install CUDA 11.8.

If you have issues about installing CUDA 11.8, please have a look at https://k2-fsa.github.io/k2/installation/cuda-cudnn.html#cuda-11-8.

Note that you don’t need to have sudo permission to install CUDA 11.8

This approach supports only Linux x64 and Windows x64.

Please use the following command to install CUDA-enabled sherpa-onnx:

# We use 1.12.23 here for demonstration.
#
# Please visit https://k2-fsa.github.io/sherpa/onnx/cuda.html
# to find available versions

pip install --verbose sherpa-onnx=="1.12.23+cuda" --no-index -f https://k2-fsa.github.io/sherpa/onnx/cuda.html

# For Chinese users, please use
# pip install --verbose sherpa-onnx=="1.12.23+cuda" --no-index -f https://k2-fsa.github.io/sherpa/onnx/cuda-cn.html

The installation logs are given below:

Looking in links: https://k2-fsa.github.io/sherpa/onnx/cuda.html
Collecting sherpa-onnx==1.12.23+cuda
  Downloading https://huggingface.co/csukuangfj2/sherpa-onnx-wheels/resolve/main/cuda/1.12.23/sherpa_onnx-1.12.23%2Bcuda-cp312-cp312-linux_x86_64.whl (190.3 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 190.3/190.3 MB 9.2 MB/s eta 0:00:00
Installing collected packages: sherpa-onnx

To check that you have installed sherpa-onnx successfully, please run:

python3 -c "import sherpa_onnx; print(sherpa_onnx.__version__)"

which should print something like below:

1.12.13+cuda

Method 3 (From pre-compiled wheels, CPU + CUDA 12.8 + CUDNN9)

pip install sherpa-onnx==1.12.13+cuda12.cudnn9 -f https://k2-fsa.github.io/sherpa/onnx/cuda.html

# For Chinese users, please use
pip install sherpa-onnx==1.12.13+cuda12.cudnn9 -f https://k2-fsa.github.io/sherpa/onnx/cuda-cn.html

The installation logs are given below:

Looking in links: https://k2-fsa.github.io/sherpa/onnx/cuda.html
Collecting sherpa-onnx==1.12.13+cuda12.cudnn9
  Downloading https://huggingface.co/csukuangfj/sherpa-onnx-wheels/resolve/main/cuda/1.12.13/sherpa_onnx-1.12.13%2Bcuda12.cudnn9-cp312-cp312-linux_x86_64.whl (245.8 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 245.8/245.8 MB 4.3 MB/s eta 0:00:00
Installing collected packages: sherpa-onnx
Successfully installed sherpa-onnx-1.12.13+cuda12.cudnn9

To check that you have installed sherpa-onnx successfully, please run:

python3 -c "import sherpa_onnx; print(sherpa_onnx.__version__)"

which should print something like below:

1.12.13+cuda12.cudnn9

Method 4 (From source)

CPUNvidia GPU (CUDA)

git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
python3 setup.py install

Method 5 (For developers)

CPUNvidia GPU (CUDA)Nvidia GPU (CUDA 12.x, CUDNN 9)

git clone https://github.com/k2-fsa/sherpa-onnx
cd sherpa-onnx
mkdir build
cd build

cmake \
  -DSHERPA_ONNX_ENABLE_PYTHON=ON \
  -DBUILD_SHARED_LIBS=ON \
  -DSHERPA_ONNX_ENABLE_CHECK=OFF \
  -DSHERPA_ONNX_ENABLE_PORTAUDIO=OFF \
  -DSHERPA_ONNX_ENABLE_C_API=OFF \
  -DSHERPA_ONNX_ENABLE_WEBSOCKET=OFF \
  ..

make -j
export PYTHONPATH=$PWD/../sherpa-onnx/python/:$PWD/lib:$PYTHONPATH
mkdir -p ../sherpa-onnx/python/sherpa_onnx/lib/
cp -v lib/_sherpa_onnx* ../sherpa-onnx/python/sherpa_onnx/lib/

Check your installation

To check that sherpa-onnx has been successfully installed, please use:

python3 -c "import sherpa_onnx; print(sherpa_onnx.__file__)"

It should print some output like below:

/Users/fangjun/py38/lib/python3.8/site-packages/sherpa_onnx/__init__.py

Please refer to:

https://github.com/k2-fsa/sherpa-onnx/tree/master/python-api-examples

for usages.

Please refer to Pre-trained models for a list of pre-trained models.