FUNASR训练
parent: 702ec03a | 补丁 | 提交 | ignore whitespace
雾聪
2023-11-07 71f3d70dd765ad3ffd6ca80c99219e8c6d72f89e 
update docs
8个文件已修改
34 ■■■■■ 已修改文件
runtime/docs/SDK_advanced_guide_offline.md 6 ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史
runtime/docs/SDK_advanced_guide_offline_en.md 6 ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史
runtime/docs/SDK_advanced_guide_offline_en_zh.md 6 ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史
runtime/docs/SDK_advanced_guide_offline_zh.md 6 ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史
runtime/docs/docker_offline_cpu_en_lists 2 ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史
runtime/docs/docker_offline_cpu_zh_lists 2 ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史
runtime/run_server.sh 4 ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史
runtime/websocket/bin/funasr-wss-server.cpp 2 ●●●●● 补丁 | 查看 | 原始文档 | blame | 历史 
 runtime/docs/SDK_advanced_guide_offline.md


@@ -57,7 +57,7 @@


  --download-model-dir /workspace/models \


  --vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx \


  --model-dir damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx  \


  --punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx \


  --punc-dir damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx \


  --lm-dir damo/speech_ngram_lm_zh-cn-ai-wesp-fst \


  --itn-dir thuduj12/fst_itn_zh > log.out 2>&1 &




@@ -78,7 +78,7 @@


  --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 \


  --punc-dir damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx \


  --itn-dir thuduj12/fst_itn_zh \


  --lm-dir damo/speech_ngram_lm_zh-cn-ai-wesp-fst \


  --decoder-thread-num 32 \


@@ -133,7 +133,7 @@


After executing the above command, the real-time speech transcription service will be started. If the model is specified as a ModelScope model id, the following models will be automatically downloaded from ModelScope:


[FSMN-VAD](https://www.modelscope.cn/models/damo/speech_fsmn_vad_zh-cn-16k-common-onnx/summary),


[Paraformer-lagre](https://www.modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx/summary),


[CT-Transformer](https://www.modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx/summary),


[CT-Transformer](https://www.modelscope.cn/models/damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx/summary),


[FST-ITN](https://www.modelscope.cn/models/thuduj12/fst_itn_zh/summary),


[Ngram lm](https://www.modelscope.cn/models/damo/speech_ngram_lm_zh-cn-ai-wesp-fst/summary)






 runtime/docs/SDK_advanced_guide_offline_en.md


@@ -58,7 +58,7 @@


  --download-model-dir /workspace/models \


  --vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx \


  --model-dir damo/speech_paraformer-large_asr_nat-en-16k-common-vocab10020-onnx  \


  --punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx > log.out 2>&1 &


  --punc-dir damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx > log.out 2>&1 &




# If you want to close ssl，please add：--certfile 0




@@ -74,7 +74,7 @@


  --download-model-dir /workspace/models \


  --model-dir damo/speech_paraformer-large_asr_nat-en-16k-common-vocab10020-onnx \


  --vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx \


  --punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx \


  --punc-dir damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx \


  --decoder-thread-num 32 \


  --io-thread-num  8 \


  --port 10095 \


@@ -125,7 +125,7 @@


After executing the above command, the real-time speech transcription service will be started. If the model is specified as a ModelScope model id, the following models will be automatically downloaded from ModelScope:


[FSMN-VAD](https://www.modelscope.cn/models/damo/speech_fsmn_vad_zh-cn-16k-common-onnx/summary),


[Paraformer-lagre](https://www.modelscope.cn/models/damo/speech_paraformer-large_asr_nat-en-16k-common-vocab10020-onnx/summary),


[CT-Transformer](https://www.modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx/summary)


[CT-Transformer](https://www.modelscope.cn/models/damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx/summary)




If you wish to deploy your fine-tuned model (e.g., 10epoch.pb), you need to manually rename the model to model.pb and replace the original model.pb in ModelScope. Then, specify the path as `model_dir`.






 runtime/docs/SDK_advanced_guide_offline_en_zh.md


@@ -49,7 +49,7 @@


  --download-model-dir /workspace/models \


  --vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx \


  --model-dir damo/speech_paraformer-large_asr_nat-en-16k-common-vocab10020-onnx  \


  --punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx  > log.out 2>&1 &


  --punc-dir damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx  > log.out 2>&1 &




# 如果您想关闭ssl，增加参数：--certfile 0




@@ -166,7 +166,7 @@


  --download-model-dir /workspace/models \


  --model-dir damo/speech_paraformer-large_asr_nat-en-16k-common-vocab10020-onnx \


  --vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx \


  --punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx \


  --punc-dir damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx \


  --decoder-thread-num 32 \


  --io-thread-num  8 \


  --port 10095 \


@@ -216,7 +216,7 @@


执行上述指令后，启动英文离线文件转写服务。如果模型指定为ModelScope中model id，会自动从MoldeScope中下载如下模型：


[FSMN-VAD模型](https://www.modelscope.cn/models/damo/speech_fsmn_vad_zh-cn-16k-common-onnx/summary),


[Paraformer-lagre模型](https://www.modelscope.cn/models/damo/speech_paraformer-large_asr_nat-en-16k-common-vocab10020-onnx/summary),


[CT-Transformer标点预测模型](https://www.modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx/summary)


[CT-Transformer标点预测模型](https://www.modelscope.cn/models/damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx/summary)




如果，您希望部署您finetune后的模型（例如10epoch.pb），需要手动将模型重命名为model.pb，并将原modelscope中模型model.pb替换掉，将路径指定为`model_dir`即可。






 runtime/docs/SDK_advanced_guide_offline_zh.md


@@ -49,7 +49,7 @@


  --download-model-dir /workspace/models \


  --vad-dir damo/speech_fsmn_vad_zh-cn-16k-common-onnx \


  --model-dir damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx  \


  --punc-dir damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx \


  --punc-dir damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx \


  --lm-dir damo/speech_ngram_lm_zh-cn-ai-wesp-fst \


  --itn-dir thuduj12/fst_itn_zh > log.out 2>&1 &




@@ -169,7 +169,7 @@


  --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 \


  --punc-dir damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx \


  --itn-dir thuduj12/fst_itn_zh \


  --decoder-thread-num 32 \


  --io-thread-num  8 \


@@ -222,7 +222,7 @@


执行上述指令后，启动离线文件转写服务。如果模型指定为ModelScope中model id，会自动从MoldeScope中下载如下模型：


[FSMN-VAD模型](https://www.modelscope.cn/models/damo/speech_fsmn_vad_zh-cn-16k-common-onnx/summary),


[Paraformer-lagre模型](https://www.modelscope.cn/models/damo/speech_paraformer-large_asr_nat-zh-cn-16k-common-vocab8404-onnx/summary),


[CT-Transformer标点预测模型](https://www.modelscope.cn/models/damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx/summary),


[CT-Transformer标点预测模型](https://www.modelscope.cn/models/damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx/summary),


[基于FST的中文ITN](https://www.modelscope.cn/models/thuduj12/fst_itn_zh/summary),


[Ngram中文语言模型](https://www.modelscope.cn/models/damo/speech_ngram_lm_zh-cn-ai-wesp-fst/summary)






 runtime/docs/docker_offline_cpu_en_lists


@@ -5,4 +5,4 @@


DEFAULT_VAD_MODEL:


  damo/speech_fsmn_vad_zh-cn-16k-common-onnx


DEFAULT_PUNC_MODEL:


  damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx


  damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx




 runtime/docs/docker_offline_cpu_zh_lists


@@ -9,7 +9,7 @@


DEFAULT_VAD_MODEL:


  damo/speech_fsmn_vad_zh-cn-16k-common-onnx


DEFAULT_PUNC_MODEL:


  damo/punc_ct-transformer_zh-cn-common-vocab272727-onnx


  damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx


DEFAULT_LM_MODEL:


  damo/speech_ngram_lm_zh-cn-ai-wesp-fst


  NONE




 runtime/run_server.sh


@@ -2,9 +2,9 @@


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"


punc_dir="damo/punc_ct-transformer_cn-en-common-vocab471067-large-onnx"


itn_dir="thuduj12/fst_itn_zh"


lm_dir=""


lm_dir="damo/speech_ngram_lm_zh-cn-ai-wesp-fst"


decoder_thread_num=32


io_thread_num=8


port=10095




 runtime/websocket/bin/funasr-wss-server.cpp


@@ -108,7 +108,7 @@


    TCLAP::ValueArg<float>    am_scale("", AM_SCALE, "the acoustic scale for beam searching ", false, 10.0, "float");




    TCLAP::ValueArg<std::string> lm_dir("", LM_DIR,


        "the LM model path, which contains compiled models: TLG.fst, config.yaml ", false, "", "string");


        "the LM model path, which contains compiled models: TLG.fst, config.yaml ", false, "damo/speech_ngram_lm_zh-cn-ai-wesp-fst", "string");


    TCLAP::ValueArg<std::string> lm_revision(


        "", "lm-revision", "LM model revision", false, "v1.0.1", "string");


    TCLAP::ValueArg<std::string> hotword("", HOTWORD,