From 8dab6d184a034ca86eafa644ea0d2100aadfe27d Mon Sep 17 00:00:00 2001 From: jmwang66 <wangjiaming.wjm@alibaba-inc.com> Date: 星期二, 09 五月 2023 10:58:33 +0800 Subject: [PATCH] Merge pull request #473 from alibaba-damo-academy/dev_smohan --- funasr/runtime/python/grpc/Readme.md | 93 ++++++++++++++++++++++++++++++++-------------- 1 files changed, 65 insertions(+), 28 deletions(-) diff --git a/funasr/runtime/python/grpc/Readme.md b/funasr/runtime/python/grpc/Readme.md index dc38b51..742268b 100644 --- a/funasr/runtime/python/grpc/Readme.md +++ b/funasr/runtime/python/grpc/Readme.md @@ -1,45 +1,82 @@ -# Using paraformer with grpc +# Service with grpc-python We can send streaming audio data to server in real-time with grpc client every 10 ms e.g., and get transcribed text when stop speaking. The audio data is in streaming, the asr inference process is in offline. +## For the Server -## Steps +### Prepare server environment +#### Backend is modelscope pipeline (default) +Install the modelscope and funasr -Step 1) Optional, prepare server environment (on server). Install modelscope and funasr with pip or with cuda-docker image. - - Option 1: Install modelscope and funasr with [pip](https://github.com/alibaba-damo-academy/FunASR#installation) - - Option 2: or install with cuda-docker image as: - -``` -CID=`docker run --network host -d -it --gpus '"device=0"' registry.cn-hangzhou.aliyuncs.com/modelscope-repo/modelscope:ubuntu20.04-cuda11.3.0-py37-torch1.11.0-tf1.15.5-1.2.0` -echo $CID -docker exec -it $CID /bin/bash -``` - Get funasr source code and get into grpc directory. -``` -git clone https://github.com/alibaba-damo-academy/FunASR -cd FunASR/funasr/runtime/python/grpc/ +```shell +pip install -U modelscope funasr +# For the users in China, you could install with the command: +# pip install -U modelscope funasr -i https://mirror.sjtu.edu.cn/pypi/web/simple +git clone https://github.com/alibaba/FunASR.git && cd FunASR ``` +Install the requirements -Step 2) Optional, generate protobuf file (run on server, the two generated pb file are both used for server and client). +```shell +cd funasr/runtime/python/grpc +pip install -r requirements_server.txt ``` -# paraformer_pb2.py and paraformer_pb2_grpc.py are already generated. + +#### Backend is funasr_onnx (optional) + +Install [`funasr_onnx`](https://github.com/alibaba-damo-academy/FunASR/tree/main/funasr/runtime/python/onnxruntime). + +``` +pip install funasr_onnx -i https://pypi.Python.org/simple +``` + +Export the model, more details ref to [export docs](https://github.com/alibaba-damo-academy/FunASR/tree/main/funasr/runtime/python/onnxruntime). +```shell +python -m funasr.export.export_model --model-name damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch --export-dir ./export --type onnx --quantize True +``` + +### Generate protobuf file +Run on server, the two generated pb files are both used for server and client + +```shell +# paraformer_pb2.py and paraformer_pb2_grpc.py are already generated, +# regenerate it only when you make changes to ./proto/paraformer.proto file. python -m grpc_tools.protoc --proto_path=./proto -I ./proto --python_out=. --grpc_python_out=./ ./proto/paraformer.proto ``` -Step 3) Start grpc server (on server). -``` -python grpc_main_server.py --port 10095 -``` - -Step 4) Start grpc client (on client with microphone). +### Start grpc server ``` -# Optional, Install dependency. -python -m pip install pyaudio webrtcvad +# Start server. +python grpc_main_server.py --port 10095 --backend pipeline ``` + +If you want run server with onnxruntime, please set `backend` and `onnx_dir`. +``` +# Start server. +python grpc_main_server.py --port 10095 --backend onnxruntime --onnx_dir /models/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch +``` + +## For the client + +### Install the requirements + +```shell +git clone https://github.com/alibaba/FunASR.git && cd FunASR +cd funasr/runtime/python/grpc +pip install -r requirements_client.txt +``` + +### Generate protobuf file +Run on server, the two generated pb files are both used for server and client + +```shell +# paraformer_pb2.py and paraformer_pb2_grpc.py are already generated, +# regenerate it only when you make changes to ./proto/paraformer.proto file. +python -m grpc_tools.protoc --proto_path=./proto -I ./proto --python_out=. --grpc_python_out=./ ./proto/paraformer.proto +``` + +### Start grpc client ``` # Start client. python grpc_main_client_mic.py --host 127.0.0.1 --port 10095 @@ -47,8 +84,8 @@ ## Workflow in desgin - +<div align="left"><img src="proto/workflow.png" width="400"/> ## Reference We borrow from or refer to some code as: -- Gitblit v1.9.1