speech_recognition/FunASR
1
0
Fork 0
mirror of https://github.com/modelscope/FunASR synced 2025-09-15 14:48:36 +08:00
Code Issues Actions Packages Projects Releases Wiki Activity

Create OnlineFeature.h

Browse Source
This commit is contained in:
zhuzizyf 2023-04-22 14:22:17 +08:00 committed by GitHub
parent f06c7f9ec0
commit 965797ecf0
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 59 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

59
funasr/runtime/onnxruntime/src/OnlineFeature.h Normal file
View File

@ -0,0 +1,59 @@
//
// Created by zhuzizyf(China Telecom Shanghai) on 4/22/23.
//
#include "kaldi-native-fbank/csrc/feature-fbank.h"
#include "kaldi-native-fbank/csrc/online-feature.h"
#include <vector>
using namespace std;
class OnlineFeature {
public:
OnlineFeature(int sample_rate, knf::FbankOptions fbank_opts, int lfr_m_, int lfr_n_,
std::vector<std::vector<float>> cmvns_);
void extractFeats(vector<vector<float>> &vad_feats, vector<float> waves, bool input_finished);
private:
void onlineFbank(vector<vector<float>> &vad_feats, vector<float> &waves);
int OnlineLfrCmvn(vector<vector<float>> &vad_feats);
static int compute_frame_num(int sample_length, int frame_sample_length, int frame_shift_sample_length) {
int frame_num = static_cast<int>((sample_length - frame_sample_length) / frame_shift_sample_length + 1);
if (frame_num >= 1 && sample_length >= frame_sample_length)
return frame_num;
else
return 0;
}
void reset_cache() {
reserve_waveforms_.clear();
input_cache_.clear();
lfr_splice_cache_.clear();
input_finished_ = false;
}
knf::FbankOptions fbank_opts_;
// The reserved waveforms by fbank
std::vector<float> reserve_waveforms_;
// waveforms reserved after last shift position
std::vector<float> input_cache_;
// lfr reserved cache
std::vector<std::vector<float>> lfr_splice_cache_;
std::vector<std::vector<float>> cmvns_;
int sample_rate_ = 16000;
int frame_sample_length_ = sample_rate_ / 1000 * 25;;
int frame_shift_sample_length_ = sample_rate_ / 1000 * 10;
int lfr_m_;
int lfr_n_;
bool input_finished_ = false;
};