From 0ff321caea4a1dcc1368f50cd0e40d199f0da7d2 Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期四, 30 五月 2024 15:10:51 +0800
Subject: [PATCH] docs

---
 runtime/websocket/bin/funasr-wss-server.cpp |   17 ++++++++++-------
 1 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/runtime/websocket/bin/funasr-wss-server.cpp b/runtime/websocket/bin/funasr-wss-server.cpp
index bd82277..100cf35 100644
--- a/runtime/websocket/bin/funasr-wss-server.cpp
+++ b/runtime/websocket/bin/funasr-wss-server.cpp
@@ -50,7 +50,7 @@
     TCLAP::ValueArg<std::string> model_revision(
         "", "model-revision",
         "ASR model revision",
-        false, "v1.2.1", "string");
+        false, "v2.0.4", "string");
     TCLAP::ValueArg<std::string> quantize(
         "", QUANTIZE,
         "true (Default), load the model of model_quant.onnx in model_dir. If set "
@@ -67,7 +67,7 @@
     TCLAP::ValueArg<std::string> vad_revision(
         "", "vad-revision",
         "VAD model revision",
-        false, "v1.2.0", "string");
+        false, "v2.0.4", "string");
     TCLAP::ValueArg<std::string> vad_quant(
         "", VAD_QUANT,
         "true (Default), load the model of model_quant.onnx in vad_dir. If set "
@@ -81,7 +81,7 @@
     TCLAP::ValueArg<std::string> punc_revision(
         "", "punc-revision",
         "PUNC model revision",
-        false, "v1.1.7", "string");
+        false, "v2.0.4", "string");
     TCLAP::ValueArg<std::string> punc_quant(
         "", PUNC_QUANT,
         "true (Default), load the model of model_quant.onnx in punc_dir. If set "
@@ -126,6 +126,7 @@
     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);
+    TCLAP::ValueArg<std::int32_t> batch_size("", BATCHSIZE, "batch_size for ASR model when using GPU", false, 4, "int32_t");
 
     // add file
     cmd.add(hotword);
@@ -158,6 +159,7 @@
     cmd.add(decoder_thread_num);
     cmd.add(model_thread_num);
     cmd.add(use_gpu);
+    cmd.add(batch_size);
     cmd.parse(argc, argv);
 
     std::map<std::string, std::string> model_path;
@@ -182,6 +184,7 @@
     lattice_beam_ = lattice_beam.getValue();
     am_scale_ = am_scale.getValue();
     bool use_gpu_ = use_gpu.getValue();
+    int batch_size_ = batch_size.getValue();
 
     // Download model form Modelscope
     try{
@@ -242,17 +245,17 @@
             // 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";
+                model_path["model-revision"]="v2.0.4";
             }
 
             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";
+                model_path["model-revision"]="v2.0.5";
             }
 
             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";
+                model_path["model-revision"]="v2.0.4";
                 s_itn_path="";
                 s_lm_path="";
             }
@@ -477,7 +480,7 @@
     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, use_gpu_);  // init asr model
+    websocket_srv.initAsr(model_path, s_model_thread_num, use_gpu_, batch_size_);  // init asr model
 
     LOG(INFO) << "decoder-thread-num: " << s_decoder_thread_num;
     LOG(INFO) << "io-thread-num: " << s_io_thread_num;

--
Gitblit v1.9.1