1、创建虚拟环境

wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh
sh Miniconda3-latest-Linux-x86_64.sh
source ~/.bashrc
conda create -n funasr python=3.7
conda activate funasr

2、安装依赖

pip3 install -U funasr -i https://mirror.sjtu.edu.cn/pypi/web/simple
pip3 install -U modelscope -f https://modelscope.oss-cn-beijing.aliyuncs.com/releases/repo.html -i https://mirror.sjtu.edu.cn/pypi/web/simple
pip3 install torch torchaudio
pip3 install onnx
pip3 install onnxruntime

python3 -m funasr.export.export_model \
--export-dir ./export \
--type onnx \
--quantize True \
--model-name damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-pytorch \
--model-name damo/speech_fsmn_vad_zh-cn-16k-common-pytorch \
--model-name damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch

3、下载onnxruntime

wget https://github.com/microsoft/onnxruntime/releases/download/v1.14.0/onnxruntime-linux-x64-1.14.0.tgz
tar -zxvf onnxruntime-linux-x64-1.14.0.tgz

4、安装系统依赖

sudo apt-get install libopenblas-dev #ubuntu
# sudo yum -y install openblas-devel #centos
apt-get install libssl-dev #ubuntu 
# yum install openssl-devel #centos

5、编译安装

cd funasr/runtime/websocket
mkdir build && cd build
cmake -DCMAKE_BUILD_TYPE=release .. -DONNXRUNTIME_DIR=/path/to/onnxruntime-linux-x64-1.14.0 -DFFMPEG_DIR=/path/to/ffmpeg-N-111383-g20b8688092-linux64-gpl-shared
make

注意：/path/to/替换为真实路径。

6、运行

参考链接：https://github.com/lukeewin/FunASR/tree/main/funasr/runtime/websocket#service-with-websocket-cpp

./funasr-wss-server 
--port 10096 
--download-model-dir /workspace/models 
--model-dir damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx 
--vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx 
--punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx 
--certfile ../../../ssl_key/server.crt 
--keyfile ../../../ssl_key/server.key 
--decoder-thread-num 16 
--io-thread-num 16

如果要加载自己的模型，把--download-model-dir去掉，把--model-dir后面的值改为你模型的路径。

7、更换模型

从魔塔社区上下载模型。

转换为onnx

python -m funasr.export.export_model \
    --model-name [model_name] \
    --export-dir [export_dir] \
    --type [onnx, torch] \
    --quantize [true, false] \
    --fallback-num [fallback_num]

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

参考文档：FunASR离线文件转写服务开发指南

8、支持时间戳

参考文档：https://github.com/alibaba-damo-academy/FunASR/discussions/246#discussioncomment-5449090

from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
param_dict = dict()
param_dict['hotword'] = "hotword.txt"
inference_pipeline = pipeline(
    task=Tasks.auto_speech_recognition,
    model="damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404",
    vad_model='damo/speech_fsmn_vad_zh-cn-16k-common-pytorch',
    punc_model='damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch',
    param_dict=param_dict)

inference_pipeline_tp = pipeline(
    task=Tasks.speech_timestamp,
    model='damo/speech_timestamp_prediction-v1-16k-offline',
    output_dir='./tmp')

rec_result_asr = inference_pipeline(audio_in="test2.wav")
rec_result_tp = inference_pipeline_tp(audio_in="test2.wav",
                                      text_in=" ".join(rec_result_asr['text']))
print(rec_result_asr, rec_result_tp)

效果：

9、支持自定义热词

模型：https://www.modelscope.cn/models/damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404/summary

from modelscope.pipelines import pipeline
from modelscope.utils.constant import Tasks
param_dict = dict()
param_dict['hotword'] = "resources/funasr/data/hotword.txt"
inference_pipeline = pipeline(
    task=Tasks.auto_speech_recognition,
    model="damo/speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404",
    vad_model='damo/speech_fsmn_vad_zh-cn-16k-common-pytorch',
    punc_model='damo/punc_ct-transformer_zh-cn-common-vocab272727-pytorch',
    param_dict=param_dict)

rec_result = inference_pipeline(audio_in='resources/funasr/audio/2.wav')
print(rec_result)

Q.E.D.