EngineX-Iluvatar/enginex_bi_series-sherpa-onnx
Archived
16
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

Add non-streaming ASR (#92)

Browse Source
This commit is contained in:
Fangjun Kuang
2023-03-26 08:53:42 +08:00
committed by GitHub
parent 6f92bc7362
commit 5572246253
48 changed files with 1526 additions and 150 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
sherpa-onnx/csrc/wave-reader.cc
View File

@@ -6,10 +6,11 @@
#include <cassert>
#include <fstream>
#include <iostream>
#include <utility>
#include <vector>
#include "sherpa-onnx/csrc/macros.h"
namespace sherpa_onnx {
namespace {
// see http://soundfile.sapp.org/doc/WaveFormat/
@@ -20,26 +21,34 @@ struct WaveHeader {
bool Validate() const {
// F F I R
if (chunk_id != 0x46464952) {
SHERPA_ONNX_LOGE("Expected chunk_id RIFF. Given: 0x%08x\n", chunk_id);
return false;
}
// E V A W
if (format != 0x45564157) {
SHERPA_ONNX_LOGE("Expected format WAVE. Given: 0x%08x\n", format);
return false;
}
if (subchunk1_id != 0x20746d66) {
SHERPA_ONNX_LOGE("Expected subchunk1_id 0x20746d66. Given: 0x%08x\n",
subchunk1_id);
return false;
}
if (subchunk1_size != 16) { // 16 for PCM
SHERPA_ONNX_LOGE("Expected subchunk1_size 16. Given: %d\n",
subchunk1_size);
return false;
}
if (audio_format != 1) { // 1 for PCM
SHERPA_ONNX_LOGE("Expected audio_format 1. Given: %d\n", audio_format);
return false;
}
if (num_channels != 1) { // we support only single channel for now
SHERPA_ONNX_LOGE("Expected single channel. Given: %d\n", num_channels);
return false;
}
if (byte_rate != (sample_rate * num_channels * bits_per_sample / 8)) {
@@ -51,6 +60,8 @@ struct WaveHeader {
}
if (bits_per_sample != 16) { // we support only 16 bits per sample
SHERPA_ONNX_LOGE("Expected bits_per_sample 16. Given: %d\n",
bits_per_sample);
return false;
}
@@ -62,7 +73,7 @@ struct WaveHeader {
// and
// https://www.robotplanet.dk/audio/wav_meta_data/riff_mci.pdf
void SeekToDataChunk(std::istream &is) {
// a t a d
// a t a d
while (is && subchunk2_id != 0x61746164) {
// const char *p = reinterpret_cast<const char *>(&subchunk2_id);
// printf("Skip chunk (%x): %c%c%c%c of size: %d\n", subchunk2_id, p[0],
@@ -91,7 +102,7 @@ static_assert(sizeof(WaveHeader) == 44, "");
// Read a wave file of mono-channel.
// Return its samples normalized to the range [-1, 1).
std::vector<float> ReadWaveImpl(std::istream &is, float expected_sample_rate,
std::vector<float> ReadWaveImpl(std::istream &is, int32_t *sampling_rate,
bool *is_ok) {
WaveHeader header;
is.read(reinterpret_cast<char *>(&header), sizeof(header));
@@ -111,10 +122,7 @@ std::vector<float> ReadWaveImpl(std::istream &is, float expected_sample_rate,
return {};
}
if (expected_sample_rate != header.sample_rate) {
*is_ok = false;
return {};
}
*sampling_rate = header.sample_rate;
// header.subchunk2_size contains the number of bytes in the data.
// As we assume each sample contains two bytes, so it is divided by 2 here
@@ -137,15 +145,15 @@ std::vector<float> ReadWaveImpl(std::istream &is, float expected_sample_rate,
} // namespace
std::vector<float> ReadWave(const std::string &filename,
float expected_sample_rate, bool *is_ok) {
std::vector<float> ReadWave(const std::string &filename, int32_t *sampling_rate,
bool *is_ok) {
std::ifstream is(filename, std::ifstream::binary);
return ReadWave(is, expected_sample_rate, is_ok);
return ReadWave(is, sampling_rate, is_ok);
}
std::vector<float> ReadWave(std::istream &is, float expected_sample_rate,
std::vector<float> ReadWave(std::istream &is, int32_t *sampling_rate,
bool *is_ok) {
auto samples = ReadWaveImpl(is, expected_sample_rate, is_ok);
auto samples = ReadWaveImpl(is, sampling_rate, is_ok);
return samples;
}