Examples: Add save audio to file option in stream.cpp (#1310)

* save the recorded audio to a file

* Alignment -help

* Save the correct audio

* chage to a consistent coding style

* Correct typo

* Update examples/stream/stream.cpp

* Update examples/stream/stream.cpp

* Correct variable misuse

* Update examples/stream/stream.cpp

* Update examples/stream/stream.cpp

* Update examples/stream/stream.cpp

* Update examples/stream/stream.cpp

---------

Co-authored-by: bobqianic <129547291+bobqianic@users.noreply.github.com>
This commit is contained in:
litong 2023-09-22 05:43:21 -10:00 committed by GitHub
parent 7e1592d2cd
commit 707507ff6d
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -2,7 +2,7 @@
//
// A very quick-n-dirty implementation serving mainly as a proof of concept.
//
#include <fstream>
#include "common-sdl.h"
#include "common.h"
#include "whisper.h"
@ -13,7 +13,60 @@
#include <thread>
#include <vector>
#include <fstream>
#include <ctime>
class SimpleWavWriter {
private:
std::ofstream file;
int32_t dataSize = 0;
public:
SimpleWavWriter(const std::string &filename, int sampleRate, int bitsPerSample, int channels) {
file.open(filename, std::ios::binary);
file.write("RIFF", 4);
file.write("\0\0\0\0", 4); // Placeholder for file size
file.write("WAVE", 4);
file.write("fmt ", 4);
int32_t subChunkSize = 16;
int16_t audioFormat = 1; // PCM format
int32_t byteRate = sampleRate * channels * bitsPerSample / 8;
int16_t blockAlign = channels * bitsPerSample / 8;
file.write(reinterpret_cast<char *>(&subChunkSize), 4);
file.write(reinterpret_cast<char *>(&audioFormat), 2);
file.write(reinterpret_cast<char *>(&channels), 2);
file.write(reinterpret_cast<char *>(&sampleRate), 4);
file.write(reinterpret_cast<char *>(&byteRate), 4);
file.write(reinterpret_cast<char *>(&blockAlign), 2);
file.write(reinterpret_cast<char *>(&bitsPerSample), 2);
file.write("data", 4);
file.write("\0\0\0\0", 4); // Placeholder for data size
}
void writeData(const float *data, size_t length) {
for (size_t i = 0; i < length; ++i) {
int16_t intSample = static_cast<int16_t>(data[i] * 32767);
file.write(reinterpret_cast<char *>(&intSample), sizeof(int16_t));
dataSize += sizeof(int16_t);
}
if (file.is_open()) {
file.seekp(4, std::ios::beg);
int32_t fileSize = 36 + dataSize;
file.write(reinterpret_cast<char *>(&fileSize), 4);
file.seekp(40, std::ios::beg);
file.write(reinterpret_cast<char *>(&dataSize), 4);
file.seekp(0, std::ios::end);
}
}
~SimpleWavWriter() {
if (file.is_open()) {
file.close();
}
}
};
// 500 -> 00:05.000
// 6000 -> 01:00.000
std::string to_timestamp(int64_t t) {
@ -52,6 +105,7 @@ struct whisper_params {
std::string language = "en";
std::string model = "models/ggml-base.en.bin";
std::string fname_out;
bool save_audio = false; // save audio to wav file
};
void whisper_print_usage(int argc, char ** argv, const whisper_params & params);
@ -82,6 +136,7 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
else if (arg == "-m" || arg == "--model") { params.model = argv[++i]; }
else if (arg == "-f" || arg == "--file") { params.fname_out = argv[++i]; }
else if (arg == "-tdrz" || arg == "--tinydiarize") { params.tinydiarize = true; }
else if (arg == "-sa" || arg == "--save-audio") { params.save_audio = true; }
else {
fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
@ -117,6 +172,7 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
fprintf(stderr, " -m FNAME, --model FNAME [%-7s] model path\n", params.model.c_str());
fprintf(stderr, " -f FNAME, --file FNAME [%-7s] text output file name\n", params.fname_out.c_str());
fprintf(stderr, " -tdrz, --tinydiarize [%-7s] enable tinydiarize (requires a tdrz model)\n", params.tinydiarize ? "true" : "false");
fprintf(stderr, " -sa, --save-audio [%-7s] save the recorded audio to a file\n", params.save_audio ? "true" : "false");
fprintf(stderr, "\n");
}
@ -154,7 +210,6 @@ int main(int argc, char ** argv) {
audio.resume();
// whisper init
if (params.language != "auto" && whisper_lang_id(params.language.c_str()) == -1){
fprintf(stderr, "error: unknown language '%s'\n", params.language.c_str());
whisper_print_usage(argc, argv, params);
@ -211,15 +266,28 @@ int main(int argc, char ** argv) {
return 1;
}
}
// save wav file
SimpleWavWriter *wavWriter = nullptr;
if (params.save_audio) {
// Get current date/time for filename
time_t now = time(0);
char buffer[80];
strftime(buffer, sizeof(buffer), "%Y%m%d%H%M%S", localtime(&now));
std::string filename = std::string(buffer) + ".wav";
printf("[Start speaking]");
wavWriter = new SimpleWavWriter(filename, WHISPER_SAMPLE_RATE, 16, 1);
}
printf("[Start speaking]\n");
fflush(stdout);
auto t_last = std::chrono::high_resolution_clock::now();
auto t_last = std::chrono::high_resolution_clock::now();
const auto t_start = t_last;
// main audio loop
while (is_running) {
if (params.save_audio && wavWriter) {
wavWriter->writeData(pcmf32_new.data(), pcmf32_new.size());
}
// handle Ctrl + C
is_running = sdl_poll_events();
@ -371,7 +439,7 @@ int main(int argc, char ** argv) {
fout << std::endl;
}
if (use_vad){
if (use_vad) {
printf("\n");
printf("### Transcription %d END\n", n_iter);
}
@ -408,4 +476,4 @@ int main(int argc, char ** argv) {
whisper_free(ctx);
return 0;
}
}