update com define

雾聪

2024-03-14 3549c0106e5a35ef2ddffdfd7381e613ed5310bd

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

@@ -4,12 +4,6 @@
 */
/* 2022-2023 by zhaomingwork */

// io server
// Usage:funasr-wss-server  [--model_thread_num <int>] [--decoder_thread_num <int>]
//                    [--io_thread_num <int>] [--port <int>] [--listen_ip
//                    <string>] [--punc-quant <string>] [--punc-dir <string>]
//                    [--vad-quant <string>] [--vad-dir <string>] [--quantize
//                    <string>] --model-dir <string> [--] [--version] [-h]
#include "websocket-server.h"
#ifdef _WIN32
#include "win_func.h"
@@ -32,12 +26,19 @@
}

int main(int argc, char* argv[]) {
#ifdef _WIN32
    #include <windows.h>
    SetConsoleOutputCP(65001);
#endif
  try {

    google::InitGoogleLogging(argv[0]);
    FLAGS_logtostderr = true;

    TCLAP::CmdLine cmd("funasr-wss-server", ' ', "1.0");
    std::string offline_version = "";
#ifdef _WIN32
    offline_version = "0.1.0";
#endif
    TCLAP::CmdLine cmd("funasr-wss-server", ' ', offline_version);
    TCLAP::ValueArg<std::string> download_model_dir(
        "", "download-model-dir",
        "Download model from Modelscope to download_model_dir",
@@ -94,7 +95,7 @@
                                           "0.0.0.0", "string");
    TCLAP::ValueArg<int> port("", "port", "port", false, 10095, "int");
    TCLAP::ValueArg<int> io_thread_num("", "io-thread-num", "io thread num",
                                       false, 8, "int");
                                       false, 2, "int");
    TCLAP::ValueArg<int> decoder_thread_num(
        "", "decoder-thread-num", "decoder thread num", false, 8, "int");
    TCLAP::ValueArg<int> model_thread_num("", "model-thread-num",
@@ -114,12 +115,13 @@
    TCLAP::ValueArg<std::string> lm_dir("", LM_DIR,
        "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");
        "", "lm-revision", "LM model revision", false, "v1.0.2", "string");
    TCLAP::ValueArg<std::string> hotword("", HOTWORD,
        "the hotword file, one hotword perline, Format: Hotword Weight (could be: 阿里巴巴 20)", 
        false, "/workspace/resources/hotwords.txt", "string");
    TCLAP::ValueArg<std::int32_t> fst_inc_wts("", FST_INC_WTS, 
        "the fst hotwords incremental bias", false, 20, "int32_t");
    TCLAP::SwitchArg use_gpu("", INFER_GPU, "Whether to use GPU, default is false", false);

    // add file
    cmd.add(hotword);
@@ -150,6 +152,7 @@
    cmd.add(io_thread_num);
    cmd.add(decoder_thread_num);
    cmd.add(model_thread_num);
    cmd.add(use_gpu);
    cmd.parse(argc, argv);

    std::map<std::string, std::string> model_path;
@@ -172,6 +175,7 @@
    global_beam_ = global_beam.getValue();
    lattice_beam_ = lattice_beam.getValue();
    am_scale_ = am_scale.getValue();
    bool use_gpu_ = use_gpu.getValue();

    // Download model form Modelscope
    try{
@@ -186,7 +190,7 @@
        std::string s_itn_path = model_path[ITN_DIR];
        std::string s_lm_path = model_path[LM_DIR];

        std::string python_cmd = "python -m funasr.utils.runtime_sdk_download_tool --type onnx --quantize True ";
        std::string python_cmd = "python -m funasr.download.runtime_sdk_download_tool --type onnx --quantize True ";

        if(vad_dir.isSet() && !s_vad_path.empty()){
            std::string python_cmd_vad;
@@ -229,28 +233,29 @@
            std::string down_asr_path;
            std::string down_asr_model;

            // modify model-revision by model name
            size_t found = s_asr_path.find("speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404");
            if (found != std::string::npos) {
                model_path["model-revision"]="v1.2.4";
            }

            found = s_asr_path.find("speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404");
            if (found != std::string::npos) {
                model_path["model-revision"]="v1.0.5";
            }

            found = s_asr_path.find("speech_paraformer-large_asr_nat-en-16k-common-vocab10020");
            if (found != std::string::npos) {
                model_path["model-revision"]="v1.0.0";
                s_itn_path="";
                s_lm_path="";
            }

            if (access(s_asr_path.c_str(), F_OK) == 0){
                // local
                python_cmd_asr = python_cmd + " --model-name " + s_asr_path + " --export-dir ./ " + " --model_revision " + model_path["model-revision"];
                down_asr_path  = s_asr_path;
            }else{
                size_t found = s_asr_path.find("speech_paraformer-large-vad-punc_asr_nat-zh-cn-16k-common-vocab8404");
                if (found != std::string::npos) {
                    model_path["model-revision"]="v1.2.4";
                }

                found = s_asr_path.find("speech_paraformer-large-contextual_asr_nat-zh-cn-16k-common-vocab8404");
                if (found != std::string::npos) {
                    model_path["model-revision"]="v1.0.5";
                }

                found = s_asr_path.find("speech_paraformer-large_asr_nat-en-16k-common-vocab10020");
                if (found != std::string::npos) {
                    model_path["model-revision"]="v1.0.0";
                    s_itn_path="";
                    s_lm_path="";
                }

                // modelscope
                LOG(INFO) << "Download model: " <<  s_asr_path << " from modelscope: ";
                python_cmd_asr = python_cmd + " --model-name " + s_asr_path + " --export-dir " + s_download_model_dir + " --model_revision " + model_path["model-revision"];
@@ -425,7 +430,7 @@
    funasr::ExtractHws(hotword_path, hws_map_);

    bool is_ssl = false;
    if (!s_certfile.empty()) {
    if (!s_certfile.empty() && access(s_certfile.c_str(), F_OK) == 0) {
      is_ssl = true;
    }

@@ -466,7 +471,7 @@
    WebSocketServer websocket_srv(
        io_decoder, is_ssl, server, wss_server, s_certfile,
        s_keyfile);  // websocket server for asr engine
    websocket_srv.initAsr(model_path, s_model_thread_num);  // init asr model
    websocket_srv.initAsr(model_path, s_model_thread_num, use_gpu_);  // init asr model

    LOG(INFO) << "decoder-thread-num: " << s_decoder_thread_num;
    LOG(INFO) << "io-thread-num: " << s_io_thread_num;