2022-10-02 16:55:45 +02:00
|
|
|
// Real-time speech recognition of input from a microphone
|
|
|
|
//
|
|
|
|
// A very quick-n-dirty implementation serving mainly as a proof of concept.
|
2022-12-15 17:28:22 +01:00
|
|
|
//
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2023-02-15 18:28:10 +01:00
|
|
|
#include "common-sdl.h"
|
2023-06-25 15:34:30 +02:00
|
|
|
#include "common.h"
|
2022-10-04 19:35:01 +02:00
|
|
|
#include "whisper.h"
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 19:35:01 +02:00
|
|
|
#include <cassert>
|
|
|
|
#include <cstdio>
|
|
|
|
#include <string>
|
|
|
|
#include <thread>
|
|
|
|
#include <vector>
|
2022-10-27 21:10:50 +02:00
|
|
|
#include <fstream>
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 19:35:01 +02:00
|
|
|
// 500 -> 00:05.000
|
|
|
|
// 6000 -> 01:00.000
|
|
|
|
std::string to_timestamp(int64_t t) {
|
|
|
|
int64_t sec = t/100;
|
|
|
|
int64_t msec = t - sec*100;
|
|
|
|
int64_t min = sec/60;
|
|
|
|
sec = sec - min*60;
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 19:35:01 +02:00
|
|
|
char buf[32];
|
|
|
|
snprintf(buf, sizeof(buf), "%02d:%02d.%03d", (int) min, (int) sec, (int) msec);
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 19:35:01 +02:00
|
|
|
return std::string(buf);
|
2022-10-02 16:55:45 +02:00
|
|
|
}
|
|
|
|
|
2022-10-04 19:35:01 +02:00
|
|
|
// command-line parameters
|
|
|
|
struct whisper_params {
|
2022-10-30 07:27:04 +01:00
|
|
|
int32_t n_threads = std::min(4, (int32_t) std::thread::hardware_concurrency());
|
|
|
|
int32_t step_ms = 3000;
|
|
|
|
int32_t length_ms = 10000;
|
2022-12-15 17:28:22 +01:00
|
|
|
int32_t keep_ms = 200;
|
2022-10-30 07:27:04 +01:00
|
|
|
int32_t capture_id = -1;
|
2022-11-20 20:22:02 +01:00
|
|
|
int32_t max_tokens = 32;
|
2022-11-20 20:12:01 +01:00
|
|
|
int32_t audio_ctx = 0;
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
float vad_thold = 0.6f;
|
|
|
|
float freq_thold = 100.0f;
|
|
|
|
|
2022-11-25 18:08:51 +01:00
|
|
|
bool speed_up = false;
|
|
|
|
bool translate = false;
|
2023-04-15 15:04:07 +02:00
|
|
|
bool no_fallback = false;
|
2022-11-25 18:08:51 +01:00
|
|
|
bool print_special = false;
|
2022-12-15 17:28:22 +01:00
|
|
|
bool no_context = true;
|
|
|
|
bool no_timestamps = false;
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 19:35:01 +02:00
|
|
|
std::string language = "en";
|
|
|
|
std::string model = "models/ggml-base.en.bin";
|
2022-12-23 19:18:51 +01:00
|
|
|
std::string fname_out;
|
2022-10-04 19:35:01 +02:00
|
|
|
};
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 19:35:01 +02:00
|
|
|
void whisper_print_usage(int argc, char ** argv, const whisper_params & params);
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 19:35:01 +02:00
|
|
|
bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
|
|
|
|
for (int i = 1; i < argc; i++) {
|
|
|
|
std::string arg = argv[i];
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-11-25 18:08:51 +01:00
|
|
|
if (arg == "-h" || arg == "--help") {
|
2022-10-04 19:35:01 +02:00
|
|
|
whisper_print_usage(argc, argv, params);
|
|
|
|
exit(0);
|
2022-11-25 18:08:51 +01:00
|
|
|
}
|
|
|
|
else if (arg == "-t" || arg == "--threads") { params.n_threads = std::stoi(argv[++i]); }
|
|
|
|
else if ( arg == "--step") { params.step_ms = std::stoi(argv[++i]); }
|
|
|
|
else if ( arg == "--length") { params.length_ms = std::stoi(argv[++i]); }
|
2022-12-15 17:28:22 +01:00
|
|
|
else if ( arg == "--keep") { params.keep_ms = std::stoi(argv[++i]); }
|
2022-11-25 18:08:51 +01:00
|
|
|
else if (arg == "-c" || arg == "--capture") { params.capture_id = std::stoi(argv[++i]); }
|
|
|
|
else if (arg == "-mt" || arg == "--max-tokens") { params.max_tokens = std::stoi(argv[++i]); }
|
|
|
|
else if (arg == "-ac" || arg == "--audio-ctx") { params.audio_ctx = std::stoi(argv[++i]); }
|
2022-12-15 17:28:22 +01:00
|
|
|
else if (arg == "-vth" || arg == "--vad-thold") { params.vad_thold = std::stof(argv[++i]); }
|
|
|
|
else if (arg == "-fth" || arg == "--freq-thold") { params.freq_thold = std::stof(argv[++i]); }
|
2022-11-25 18:08:51 +01:00
|
|
|
else if (arg == "-su" || arg == "--speed-up") { params.speed_up = true; }
|
|
|
|
else if (arg == "-tr" || arg == "--translate") { params.translate = true; }
|
2023-04-15 15:04:07 +02:00
|
|
|
else if (arg == "-nf" || arg == "--no-fallback") { params.no_fallback = true; }
|
2022-11-25 18:08:51 +01:00
|
|
|
else if (arg == "-ps" || arg == "--print-special") { params.print_special = true; }
|
2022-12-15 17:28:22 +01:00
|
|
|
else if (arg == "-kc" || arg == "--keep-context") { params.no_context = false; }
|
2022-11-25 18:08:51 +01:00
|
|
|
else if (arg == "-l" || arg == "--language") { params.language = argv[++i]; }
|
|
|
|
else if (arg == "-m" || arg == "--model") { params.model = argv[++i]; }
|
|
|
|
else if (arg == "-f" || arg == "--file") { params.fname_out = argv[++i]; }
|
|
|
|
else {
|
2022-10-04 19:35:01 +02:00
|
|
|
fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
|
|
|
|
whisper_print_usage(argc, argv, params);
|
|
|
|
exit(0);
|
2022-10-02 16:55:45 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2022-12-19 19:45:08 +01:00
|
|
|
void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & params) {
|
2022-10-04 19:35:01 +02:00
|
|
|
fprintf(stderr, "\n");
|
|
|
|
fprintf(stderr, "usage: %s [options]\n", argv[0]);
|
|
|
|
fprintf(stderr, "\n");
|
|
|
|
fprintf(stderr, "options:\n");
|
2022-11-25 18:08:51 +01:00
|
|
|
fprintf(stderr, " -h, --help [default] show this help message and exit\n");
|
2023-04-15 15:04:07 +02:00
|
|
|
fprintf(stderr, " -t N, --threads N [%-7d] number of threads to use during computation\n", params.n_threads);
|
|
|
|
fprintf(stderr, " --step N [%-7d] audio step size in milliseconds\n", params.step_ms);
|
|
|
|
fprintf(stderr, " --length N [%-7d] audio length in milliseconds\n", params.length_ms);
|
|
|
|
fprintf(stderr, " --keep N [%-7d] audio to keep from previous step in ms\n", params.keep_ms);
|
|
|
|
fprintf(stderr, " -c ID, --capture ID [%-7d] capture device ID\n", params.capture_id);
|
|
|
|
fprintf(stderr, " -mt N, --max-tokens N [%-7d] maximum number of tokens per audio chunk\n", params.max_tokens);
|
|
|
|
fprintf(stderr, " -ac N, --audio-ctx N [%-7d] audio context size (0 - all)\n", params.audio_ctx);
|
|
|
|
fprintf(stderr, " -vth N, --vad-thold N [%-7.2f] voice activity detection threshold\n", params.vad_thold);
|
|
|
|
fprintf(stderr, " -fth N, --freq-thold N [%-7.2f] high-pass frequency cutoff\n", params.freq_thold);
|
|
|
|
fprintf(stderr, " -su, --speed-up [%-7s] speed up audio by x2 (reduced accuracy)\n", params.speed_up ? "true" : "false");
|
|
|
|
fprintf(stderr, " -tr, --translate [%-7s] translate from source language to english\n", params.translate ? "true" : "false");
|
|
|
|
fprintf(stderr, " -nf, --no-fallback [%-7s] do not use temperature fallback while decoding\n", params.no_fallback ? "true" : "false");
|
|
|
|
fprintf(stderr, " -ps, --print-special [%-7s] print special tokens\n", params.print_special ? "true" : "false");
|
|
|
|
fprintf(stderr, " -kc, --keep-context [%-7s] keep context between audio chunks\n", params.no_context ? "false" : "true");
|
|
|
|
fprintf(stderr, " -l LANG, --language LANG [%-7s] spoken language\n", params.language.c_str());
|
|
|
|
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());
|
2022-10-04 19:35:01 +02:00
|
|
|
fprintf(stderr, "\n");
|
2022-10-02 16:55:45 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
int main(int argc, char ** argv) {
|
|
|
|
whisper_params params;
|
|
|
|
|
|
|
|
if (whisper_params_parse(argc, argv, params) == false) {
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2023-01-18 20:22:52 +01:00
|
|
|
params.keep_ms = std::min(params.keep_ms, params.step_ms);
|
|
|
|
params.length_ms = std::max(params.length_ms, params.step_ms);
|
2022-12-15 17:28:22 +01:00
|
|
|
|
2023-02-15 18:28:10 +01:00
|
|
|
const int n_samples_step = (1e-3*params.step_ms )*WHISPER_SAMPLE_RATE;
|
|
|
|
const int n_samples_len = (1e-3*params.length_ms)*WHISPER_SAMPLE_RATE;
|
|
|
|
const int n_samples_keep = (1e-3*params.keep_ms )*WHISPER_SAMPLE_RATE;
|
|
|
|
const int n_samples_30s = (1e-3*30000.0 )*WHISPER_SAMPLE_RATE;
|
2022-12-15 17:28:22 +01:00
|
|
|
|
2022-12-16 17:04:19 +01:00
|
|
|
const bool use_vad = n_samples_step <= 0; // sliding window mode uses VAD
|
2022-12-15 17:28:22 +01:00
|
|
|
|
2023-01-18 20:22:52 +01:00
|
|
|
const int n_new_line = !use_vad ? std::max(1, params.length_ms / params.step_ms - 1) : 1; // number of steps to print new line
|
2023-01-02 09:20:50 +01:00
|
|
|
|
2023-01-16 18:37:06 +01:00
|
|
|
params.no_timestamps = !use_vad;
|
|
|
|
params.no_context |= use_vad;
|
|
|
|
params.max_tokens = 0;
|
2022-12-15 17:28:22 +01:00
|
|
|
|
2022-10-02 16:55:45 +02:00
|
|
|
// init audio
|
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
audio_async audio(params.length_ms);
|
|
|
|
if (!audio.init(params.capture_id, WHISPER_SAMPLE_RATE)) {
|
|
|
|
fprintf(stderr, "%s: audio.init() failed!\n", __func__);
|
2022-10-02 16:55:45 +02:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
audio.resume();
|
|
|
|
|
|
|
|
// whisper init
|
|
|
|
|
2023-04-14 19:02:18 +02:00
|
|
|
if (params.language != "auto" && whisper_lang_id(params.language.c_str()) == -1){
|
2022-11-25 18:08:51 +01:00
|
|
|
fprintf(stderr, "error: unknown language '%s'\n", params.language.c_str());
|
|
|
|
whisper_print_usage(argc, argv, params);
|
|
|
|
exit(0);
|
|
|
|
}
|
|
|
|
|
2023-01-08 12:03:33 +01:00
|
|
|
struct whisper_context * ctx = whisper_init_from_file(params.model.c_str());
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
std::vector<float> pcmf32 (n_samples_30s, 0.0f);
|
2023-01-14 18:20:47 +01:00
|
|
|
std::vector<float> pcmf32_old;
|
2022-12-15 17:28:22 +01:00
|
|
|
std::vector<float> pcmf32_new(n_samples_30s, 0.0f);
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-11-22 17:10:35 +01:00
|
|
|
std::vector<whisper_token> prompt_tokens;
|
2022-10-18 19:15:06 +02:00
|
|
|
|
2022-10-02 16:55:45 +02:00
|
|
|
// print some info about the processing
|
|
|
|
{
|
2022-10-27 21:10:50 +02:00
|
|
|
fprintf(stderr, "\n");
|
2022-10-04 19:35:01 +02:00
|
|
|
if (!whisper_is_multilingual(ctx)) {
|
2022-10-02 16:55:45 +02:00
|
|
|
if (params.language != "en" || params.translate) {
|
|
|
|
params.language = "en";
|
|
|
|
params.translate = false;
|
2022-10-27 21:10:50 +02:00
|
|
|
fprintf(stderr, "%s: WARNING: model is not multilingual, ignoring language and translation options\n", __func__);
|
2022-10-02 16:55:45 +02:00
|
|
|
}
|
|
|
|
}
|
2022-12-15 17:28:22 +01:00
|
|
|
fprintf(stderr, "%s: processing %d samples (step = %.1f sec / len = %.1f sec / keep = %.1f sec), %d threads, lang = %s, task = %s, timestamps = %d ...\n",
|
2022-10-10 21:06:18 +02:00
|
|
|
__func__,
|
2022-12-15 17:28:22 +01:00
|
|
|
n_samples_step,
|
|
|
|
float(n_samples_step)/WHISPER_SAMPLE_RATE,
|
|
|
|
float(n_samples_len )/WHISPER_SAMPLE_RATE,
|
|
|
|
float(n_samples_keep)/WHISPER_SAMPLE_RATE,
|
2022-10-10 21:06:18 +02:00
|
|
|
params.n_threads,
|
2022-10-04 19:35:01 +02:00
|
|
|
params.language.c_str(),
|
2022-10-02 16:55:45 +02:00
|
|
|
params.translate ? "translate" : "transcribe",
|
|
|
|
params.no_timestamps ? 0 : 1);
|
2022-10-18 19:15:06 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
if (!use_vad) {
|
2023-01-16 18:37:06 +01:00
|
|
|
fprintf(stderr, "%s: n_new_line = %d, no_context = %d\n", __func__, n_new_line, params.no_context);
|
2022-12-15 17:28:22 +01:00
|
|
|
} else {
|
|
|
|
fprintf(stderr, "%s: using VAD, will transcribe on speech activity\n", __func__);
|
|
|
|
}
|
|
|
|
|
2022-10-27 21:10:50 +02:00
|
|
|
fprintf(stderr, "\n");
|
2022-10-02 16:55:45 +02:00
|
|
|
}
|
|
|
|
|
2022-10-10 21:06:18 +02:00
|
|
|
int n_iter = 0;
|
2022-12-15 17:28:22 +01:00
|
|
|
|
2022-10-02 19:11:17 +02:00
|
|
|
bool is_running = true;
|
|
|
|
|
2022-10-27 21:10:50 +02:00
|
|
|
std::ofstream fout;
|
|
|
|
if (params.fname_out.length() > 0) {
|
|
|
|
fout.open(params.fname_out);
|
|
|
|
if (!fout.is_open()) {
|
|
|
|
fprintf(stderr, "%s: failed to open output file '%s'!\n", __func__, params.fname_out.c_str());
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
printf("[Start speaking]");
|
|
|
|
fflush(stdout);
|
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
auto t_last = std::chrono::high_resolution_clock::now();
|
|
|
|
const auto t_start = t_last;
|
|
|
|
|
2022-10-02 16:55:45 +02:00
|
|
|
// main audio loop
|
2022-10-02 19:11:17 +02:00
|
|
|
while (is_running) {
|
2022-11-25 18:08:51 +01:00
|
|
|
// handle Ctrl + C
|
2023-02-15 18:28:10 +01:00
|
|
|
is_running = sdl_poll_events();
|
2022-10-02 19:11:17 +02:00
|
|
|
|
2022-10-27 21:10:50 +02:00
|
|
|
if (!is_running) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2022-10-18 19:15:06 +02:00
|
|
|
// process new audio
|
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
if (!use_vad) {
|
|
|
|
while (true) {
|
|
|
|
audio.get(params.step_ms, pcmf32_new);
|
2022-10-18 19:15:06 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
if ((int) pcmf32_new.size() > 2*n_samples_step) {
|
|
|
|
fprintf(stderr, "\n\n%s: WARNING: cannot process audio fast enough, dropping audio ...\n\n", __func__);
|
|
|
|
audio.clear();
|
|
|
|
continue;
|
|
|
|
}
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
if ((int) pcmf32_new.size() >= n_samples_step) {
|
|
|
|
audio.clear();
|
|
|
|
break;
|
|
|
|
}
|
2022-10-10 21:06:18 +02:00
|
|
|
|
2023-02-15 18:28:10 +01:00
|
|
|
std::this_thread::sleep_for(std::chrono::milliseconds(1));
|
2022-12-15 17:28:22 +01:00
|
|
|
}
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
const int n_samples_new = pcmf32_new.size();
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
// take up to params.length_ms audio from previous iteration
|
|
|
|
const int n_samples_take = std::min((int) pcmf32_old.size(), std::max(0, n_samples_keep + n_samples_len - n_samples_new));
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
//printf("processing: take = %d, new = %d, old = %d\n", n_samples_take, n_samples_new, (int) pcmf32_old.size());
|
|
|
|
|
|
|
|
pcmf32.resize(n_samples_new + n_samples_take);
|
|
|
|
|
|
|
|
for (int i = 0; i < n_samples_take; i++) {
|
|
|
|
pcmf32[i] = pcmf32_old[pcmf32_old.size() - n_samples_take + i];
|
|
|
|
}
|
|
|
|
|
|
|
|
memcpy(pcmf32.data() + n_samples_take, pcmf32_new.data(), n_samples_new*sizeof(float));
|
|
|
|
|
|
|
|
pcmf32_old = pcmf32;
|
|
|
|
} else {
|
|
|
|
const auto t_now = std::chrono::high_resolution_clock::now();
|
|
|
|
const auto t_diff = std::chrono::duration_cast<std::chrono::milliseconds>(t_now - t_last).count();
|
|
|
|
|
|
|
|
if (t_diff < 2000) {
|
|
|
|
std::this_thread::sleep_for(std::chrono::milliseconds(100));
|
|
|
|
|
|
|
|
continue;
|
|
|
|
}
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
audio.get(2000, pcmf32_new);
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2023-02-15 18:28:10 +01:00
|
|
|
if (::vad_simple(pcmf32_new, WHISPER_SAMPLE_RATE, 1000, params.vad_thold, params.freq_thold, false)) {
|
2022-12-15 17:28:22 +01:00
|
|
|
audio.get(params.length_ms, pcmf32);
|
|
|
|
} else {
|
|
|
|
std::this_thread::sleep_for(std::chrono::milliseconds(100));
|
|
|
|
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
t_last = t_now;
|
|
|
|
}
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 21:43:37 +02:00
|
|
|
// run the inference
|
|
|
|
{
|
2022-10-18 17:17:24 +02:00
|
|
|
whisper_full_params wparams = whisper_full_default_params(WHISPER_SAMPLING_GREEDY);
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-11-25 18:08:51 +01:00
|
|
|
wparams.print_progress = false;
|
|
|
|
wparams.print_special = params.print_special;
|
|
|
|
wparams.print_realtime = false;
|
|
|
|
wparams.print_timestamps = !params.no_timestamps;
|
|
|
|
wparams.translate = params.translate;
|
2022-12-15 17:28:22 +01:00
|
|
|
wparams.single_segment = !use_vad;
|
2022-11-25 18:08:51 +01:00
|
|
|
wparams.max_tokens = params.max_tokens;
|
|
|
|
wparams.language = params.language.c_str();
|
|
|
|
wparams.n_threads = params.n_threads;
|
|
|
|
|
|
|
|
wparams.audio_ctx = params.audio_ctx;
|
|
|
|
wparams.speed_up = params.speed_up;
|
|
|
|
|
2023-01-15 10:29:57 +01:00
|
|
|
// disable temperature fallback
|
2023-04-15 15:04:07 +02:00
|
|
|
//wparams.temperature_inc = -1.0f;
|
|
|
|
wparams.temperature_inc = params.no_fallback ? 0.0f : wparams.temperature_inc;
|
2023-01-15 10:29:57 +01:00
|
|
|
|
2022-11-25 18:08:51 +01:00
|
|
|
wparams.prompt_tokens = params.no_context ? nullptr : prompt_tokens.data();
|
|
|
|
wparams.prompt_n_tokens = params.no_context ? 0 : prompt_tokens.size();
|
2022-11-22 17:20:05 +01:00
|
|
|
|
2022-10-04 21:43:37 +02:00
|
|
|
if (whisper_full(ctx, wparams, pcmf32.data(), pcmf32.size()) != 0) {
|
|
|
|
fprintf(stderr, "%s: failed to process audio\n", argv[0]);
|
|
|
|
return 6;
|
2022-10-02 16:55:45 +02:00
|
|
|
}
|
|
|
|
|
2022-10-04 21:43:37 +02:00
|
|
|
// print result;
|
|
|
|
{
|
2022-12-15 17:28:22 +01:00
|
|
|
if (!use_vad) {
|
|
|
|
printf("\33[2K\r");
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
// print long empty line to clear the previous line
|
|
|
|
printf("%s", std::string(100, ' ').c_str());
|
2022-10-27 21:10:50 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
printf("\33[2K\r");
|
|
|
|
} else {
|
|
|
|
const int64_t t1 = (t_last - t_start).count()/1000000;
|
|
|
|
const int64_t t0 = std::max(0.0, t1 - pcmf32.size()*1000.0/WHISPER_SAMPLE_RATE);
|
|
|
|
|
|
|
|
printf("\n");
|
2022-12-15 19:15:36 +01:00
|
|
|
printf("### Transcription %d START | t0 = %d ms | t1 = %d ms\n", n_iter, (int) t0, (int) t1);
|
2022-12-15 17:28:22 +01:00
|
|
|
printf("\n");
|
|
|
|
}
|
2022-10-27 21:10:50 +02:00
|
|
|
|
2022-10-04 21:43:37 +02:00
|
|
|
const int n_segments = whisper_full_n_segments(ctx);
|
|
|
|
for (int i = 0; i < n_segments; ++i) {
|
|
|
|
const char * text = whisper_full_get_segment_text(ctx, i);
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 21:43:37 +02:00
|
|
|
if (params.no_timestamps) {
|
2022-10-27 21:10:50 +02:00
|
|
|
printf("%s", text);
|
2022-10-04 21:43:37 +02:00
|
|
|
fflush(stdout);
|
2022-10-27 21:10:50 +02:00
|
|
|
|
|
|
|
if (params.fname_out.length() > 0) {
|
|
|
|
fout << text;
|
|
|
|
}
|
2022-10-02 16:55:45 +02:00
|
|
|
} else {
|
2022-10-04 21:43:37 +02:00
|
|
|
const int64_t t0 = whisper_full_get_segment_t0(ctx, i);
|
|
|
|
const int64_t t1 = whisper_full_get_segment_t1(ctx, i);
|
2022-10-02 16:55:45 +02:00
|
|
|
|
2022-10-04 21:43:37 +02:00
|
|
|
printf ("[%s --> %s] %s\n", to_timestamp(t0).c_str(), to_timestamp(t1).c_str(), text);
|
2022-10-27 21:10:50 +02:00
|
|
|
|
|
|
|
if (params.fname_out.length() > 0) {
|
|
|
|
fout << "[" << to_timestamp(t0) << " --> " << to_timestamp(t1) << "] " << text << std::endl;
|
|
|
|
}
|
2022-10-04 21:43:37 +02:00
|
|
|
}
|
2022-10-02 16:55:45 +02:00
|
|
|
}
|
2022-10-27 21:10:50 +02:00
|
|
|
|
|
|
|
if (params.fname_out.length() > 0) {
|
|
|
|
fout << std::endl;
|
|
|
|
}
|
2022-12-15 17:28:22 +01:00
|
|
|
|
|
|
|
if (use_vad){
|
|
|
|
printf("\n");
|
|
|
|
printf("### Transcription %d END\n", n_iter);
|
|
|
|
}
|
2022-10-02 16:55:45 +02:00
|
|
|
}
|
2022-10-10 21:06:18 +02:00
|
|
|
|
|
|
|
++n_iter;
|
2022-10-18 19:15:06 +02:00
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
if (!use_vad && (n_iter % n_new_line) == 0) {
|
2022-10-10 21:06:18 +02:00
|
|
|
printf("\n");
|
|
|
|
|
2022-11-11 21:33:10 +01:00
|
|
|
// keep part of the audio for next iteration to try to mitigate word boundary issues
|
|
|
|
pcmf32_old = std::vector<float>(pcmf32.end() - n_samples_keep, pcmf32.end());
|
2022-11-22 17:10:35 +01:00
|
|
|
|
|
|
|
// Add tokens of the last full length segment as the prompt
|
2022-11-22 17:20:05 +01:00
|
|
|
if (!params.no_context) {
|
|
|
|
prompt_tokens.clear();
|
|
|
|
|
|
|
|
const int n_segments = whisper_full_n_segments(ctx);
|
|
|
|
for (int i = 0; i < n_segments; ++i) {
|
|
|
|
const int token_count = whisper_full_n_tokens(ctx, i);
|
|
|
|
for (int j = 0; j < token_count; ++j) {
|
|
|
|
prompt_tokens.push_back(whisper_full_get_token_id(ctx, i, j));
|
|
|
|
}
|
2022-11-22 17:10:35 +01:00
|
|
|
}
|
|
|
|
}
|
2022-10-10 21:06:18 +02:00
|
|
|
}
|
2023-04-23 16:00:30 +02:00
|
|
|
fflush(stdout);
|
2022-10-02 16:55:45 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2022-12-15 17:28:22 +01:00
|
|
|
audio.pause();
|
2022-11-25 22:07:42 +01:00
|
|
|
|
2022-10-04 19:35:01 +02:00
|
|
|
whisper_print_timings(ctx);
|
|
|
|
whisper_free(ctx);
|
2022-10-02 16:55:45 +02:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|