Merge branch 'main' of github.com:alibaba-damo-academy/FunASR add

游雁

2023-11-09 6985a4626bb1034a80e08609a313913d2df48a29

Merge branch 'main' of github.com:alibaba-damo-academy/FunASR
add

 README.md

@@ -17,7 +17,7 @@
| [**Quick Start**](#quick-start)
| [**Runtime**](./runtime/readme.md)
| [**Model Zoo**](#model-zoo)
| [**Contact**](#community-communication)
| [**Contact**](#contact)


<a name="highlights"></a>
@@ -126,7 +126,7 @@
For more detailed information, please refer to the [service deployment documentation](runtime/readme.md).


<a name="Community Communication"></a>
<a name="contact"></a>
## Community Communication
If you encounter problems in use, you can directly raise Issues on the github page.


 funasr/bin/build_trainer.py

@@ -548,7 +548,10 @@
    init_param = modelscope_dict['init_model']
    cmvn_file = modelscope_dict['cmvn_file']
    seg_dict_file = modelscope_dict['seg_dict']
    if 'bpemodel' in modelscope_dict:
    bpemodel = modelscope_dict['bpemodel']
    else:
        bpemodel = None

    # overwrite parameters
    with open(config) as f:
@@ -582,7 +585,7 @@
        args.seg_dict_file = seg_dict_file
    else:
        args.seg_dict_file = None
    if os.path.exists(bpemodel):
    if bpemodel is not None and os.path.exists(bpemodel):
        args.bpemodel = bpemodel
    else:
        args.bpemodel = None

 runtime/onnxruntime/bin/funasr-onnx-2pass-rtf.cpp

@@ -279,7 +279,7 @@
    // hotword file
    unordered_map<string, int> hws_map;
    std::string nn_hotwords_ = "";
    std::string hotword_path = model_path.at(HOTWORD);
    std::string hotword_path = hotword.getValue();
    LOG(INFO) << "hotword path: " << hotword_path;
    funasr::ExtractHws(hotword_path, hws_map, nn_hotwords_);


 runtime/onnxruntime/bin/funasr-onnx-2pass.cpp

@@ -113,7 +113,7 @@
    // hotword file
    unordered_map<string, int> hws_map;
    std::string nn_hotwords_ = "";
    std::string hotword_path = model_path.at(HOTWORD);
    std::string hotword_path = hotword.getValue();
    LOG(INFO) << "hotword path: " << hotword_path;
    funasr::ExtractHws(hotword_path, hws_map, nn_hotwords_);


 runtime/onnxruntime/src/tokenizer.cpp

@@ -17,9 +17,13 @@

CTokenizer::~CTokenizer()
{
    if (jieba_dict_trie_){
    delete jieba_dict_trie_;
    }
    if (jieba_model_){
    delete jieba_model_;
}
}

void CTokenizer::SetJiebaRes(cppjieba::DictTrie *dict, cppjieba::HMMModel *hmm) {
    jieba_processor_.SetJiebaRes(dict, hmm);

 runtime/onnxruntime/src/tokenizer.h

@@ -17,8 +17,8 @@
    vector<string>   m_id2token,m_id2punc;
    map<string, int>  m_token2id,m_punc2id;

    cppjieba::DictTrie *jieba_dict_trie_;
    cppjieba::HMMModel *jieba_model_;
    cppjieba::DictTrie *jieba_dict_trie_=nullptr;
    cppjieba::HMMModel *jieba_model_=nullptr;
    cppjieba::Jieba jieba_processor_;

public: