From de87e1d180d214e1f49682d2b5fb7c9d2c89ae7e Mon Sep 17 00:00:00 2001
From: 雾聪 <wucong.lyb@alibaba-inc.com>
Date: 星期三, 13 十二月 2023 13:57:59 +0800
Subject: [PATCH] adapted pcm to 8k
---
runtime/onnxruntime/bin/funasr-onnx-offline-rtf.cpp | 10 ++++++----
1 files changed, 6 insertions(+), 4 deletions(-)
diff --git a/runtime/onnxruntime/bin/funasr-onnx-offline-rtf.cpp b/runtime/onnxruntime/bin/funasr-onnx-offline-rtf.cpp
index b248bca..39928b4 100644
--- a/runtime/onnxruntime/bin/funasr-onnx-offline-rtf.cpp
+++ b/runtime/onnxruntime/bin/funasr-onnx-offline-rtf.cpp
@@ -29,7 +29,7 @@
std::atomic<int> wav_index(0);
std::mutex mtx;
-void runReg(FUNASR_HANDLE asr_handle, vector<string> wav_list, vector<string> wav_ids,
+void runReg(FUNASR_HANDLE asr_handle, vector<string> wav_list, vector<string> wav_ids, int audio_fs,
float* total_length, long* total_time, int core_id, float glob_beam = 3.0f, float lat_beam = 3.0f, float am_sc = 10.0f,
int fst_inc_wts = 20, string hotword_path = "") {
@@ -55,7 +55,7 @@
for (size_t i = 0; i < 1; i++)
{
FunOfflineReset(asr_handle, decoder_handle);
- FUNASR_RESULT result=FunOfflineInfer(asr_handle, wav_list[0].c_str(), RASR_NONE, NULL, hotwords_embedding, 16000, true, decoder_handle);
+ FUNASR_RESULT result=FunOfflineInfer(asr_handle, wav_list[0].c_str(), RASR_NONE, NULL, hotwords_embedding, audio_fs, true, decoder_handle);
if(result){
FunASRFreeResult(result);
}
@@ -69,7 +69,7 @@
}
gettimeofday(&start, NULL);
- FUNASR_RESULT result=FunOfflineInfer(asr_handle, wav_list[i].c_str(), RASR_NONE, NULL, hotwords_embedding, 16000, true, decoder_handle);
+ FUNASR_RESULT result=FunOfflineInfer(asr_handle, wav_list[i].c_str(), RASR_NONE, NULL, hotwords_embedding, audio_fs, true, decoder_handle);
gettimeofday(&end, NULL);
seconds = (end.tv_sec - start.tv_sec);
@@ -138,6 +138,7 @@
TCLAP::ValueArg<std::string> itn_dir("", ITN_DIR, "the itn model(fst) path, which contains zh_itn_tagger.fst and zh_itn_verbalizer.fst", false, "", "string");
TCLAP::ValueArg<std::string> wav_path("", WAV_PATH, "the input could be: wav_path, e.g.: asr_example.wav; pcm_path, e.g.: asr_example.pcm; wav.scp, kaldi style wav list (wav_id \t wav_path)", true, "", "string");
+ TCLAP::ValueArg<std::int32_t> audio_fs("", AUDIO_FS, "the sample rate of audio", false, 16000, "int32_t");
TCLAP::ValueArg<std::int32_t> thread_num("", THREAD_NUM, "multi-thread num for rtf", true, 0, "int32_t");
TCLAP::ValueArg<std::string> hotword("", HOTWORD, "the hotword file, one hotword perline, Format: Hotword Weight (could be: 闃块噷宸村反 20)", false, "", "string");
@@ -155,6 +156,7 @@
cmd.add(hotword);
cmd.add(fst_inc_wts);
cmd.add(wav_path);
+ cmd.add(audio_fs);
cmd.add(thread_num);
cmd.parse(argc, argv);
@@ -234,7 +236,7 @@
}
for (int i = 0; i < rtf_threds; i++)
{
- threads.emplace_back(thread(runReg, asr_handle, wav_list, wav_ids, &total_length, &total_time, i, glob_beam, lat_beam, am_sc, value_bias, hotword_path));
+ threads.emplace_back(thread(runReg, asr_handle, wav_list, wav_ids, audio_fs.getValue(), &total_length, &total_time, i, glob_beam, lat_beam, am_sc, value_bias, hotword_path));
}
for (auto& thread : threads)
--
Gitblit v1.9.1