ExternalVendorCode/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2024-12-19 04:37:51 +00:00
Code Issues Actions 4 Packages Projects Releases Wiki Activity

stream : add "max_tokens" cli arg

Browse Source 
Controls the max tokens per segment for the stream example
This commit is contained in:
Georgi Gerganov 2022-11-20 21:22:02 +02:00
parent fb8d77f760
commit f2df9bd768
1 changed files with 5 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
examples/stream/stream.cpp
View File

@ -40,6 +40,7 @@ struct whisper_params {
int32_t step_ms = 3000; int32_t step_ms = 3000;
int32_t length_ms = 10000; int32_t length_ms = 10000;
int32_t capture_id = -1; int32_t capture_id = -1;
int32_t max_tokens = 32;
int32_t audio_ctx = 0; int32_t audio_ctx = 0;
bool speed_up = false; bool speed_up = false;
@ -70,6 +71,8 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
params.length_ms = std::stoi(argv[++i]); params.length_ms = std::stoi(argv[++i]);
} else if (arg == "-c" || arg == "--capture") { } else if (arg == "-c" || arg == "--capture") {
params.capture_id = std::stoi(argv[++i]); 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") { } else if (arg == "-ac" || arg == "--audio_ctx") {
params.audio_ctx = std::stoi(argv[++i]); params.audio_ctx = std::stoi(argv[++i]);
} else if (arg == "-su" || arg == "--speed-up") { } else if (arg == "-su" || arg == "--speed-up") {
@ -119,6 +122,7 @@ void whisper_print_usage(int argc, char ** argv, const whisper_params & params)
fprintf(stderr, " --step N audio step size in milliseconds (default: %d)\n", params.step_ms); fprintf(stderr, " --step N audio step size in milliseconds (default: %d)\n", params.step_ms);
fprintf(stderr, " --length N audio length in milliseconds (default: %d)\n", params.length_ms); fprintf(stderr, " --length N audio length in milliseconds (default: %d)\n", params.length_ms);
fprintf(stderr, " -c ID, --capture ID capture device ID (default: -1)\n"); fprintf(stderr, " -c ID, --capture ID capture device ID (default: -1)\n");
fprintf(stderr, " -mt N, --max_tokens N maximum number of tokens per audio chunk (default: %d)\n", params.max_tokens);
fprintf(stderr, " -ac N, --audio_ctx N audio context size (default: %d, 0 - all)\n", params.audio_ctx); fprintf(stderr, " -ac N, --audio_ctx N audio context size (default: %d, 0 - all)\n", params.audio_ctx);
fprintf(stderr, " -su, --speed-up speed up audio by factor of 2 (faster processing, reduced accuracy, default: %s)\n", params.speed_up ? "true" : "false"); fprintf(stderr, " -su, --speed-up speed up audio by factor of 2 (faster processing, reduced accuracy, default: %s)\n", params.speed_up ? "true" : "false");
fprintf(stderr, " -v, --verbose verbose output\n"); fprintf(stderr, " -v, --verbose verbose output\n");
@ -333,7 +337,7 @@ int main(int argc, char ** argv) {
wparams.translate = params.translate; wparams.translate = params.translate;
wparams.no_context = params.no_context; wparams.no_context = params.no_context;
wparams.single_segment = true; wparams.single_segment = true;
wparams.max_tokens = 32; wparams.max_tokens = params.max_tokens;
wparams.language = params.language.c_str(); wparams.language = params.language.c_str();
wparams.n_threads = params.n_threads; wparams.n_threads = params.n_threads;