ExternalVendorCode/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2025-06-12 20:18:08 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

ggml : sync latest ggml lib

Browse Source
This commit is contained in:
Georgi Gerganov
2023-06-25 14:22:21 +03:00
parent 7dfc11843c
commit 5feb0dffba
11 changed files with 6948 additions and 1158 deletions
Download Patch File Download Diff File Expand all files Collapse all files

69
examples/main/main.cpp
View File

@ -10,6 +10,10 @@
#include <vector>
#include <cstring>
#if defined(_MSC_VER)
#pragma warning(disable: 4244 4267) // possible loss of data
#endif
// Terminal color map. 10 colors grouped in ranges [0.0, 0.1, ..., 0.9]
// Lowest is red, middle is yellow, highest is green.
const std::vector<std::string> k_colors = {
@ -148,7 +152,8 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
else if (arg == "-f" || arg == "--file") { params.fname_inp.emplace_back(argv[++i]); }
else {
fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
return false;
whisper_print_usage(argc, argv, params);
exit(0);
}
}
@ -423,13 +428,13 @@ bool output_json(struct whisper_context * ctx, const char * fname, const whisper
indent++;
};
auto end_arr = [&](bool end = false) {
auto end_arr = [&](bool end) {
indent--;
doindent();
fout << (end ? "]\n" : "},\n");
};
auto start_obj = [&](const char *name = nullptr) {
auto start_obj = [&](const char *name) {
doindent();
if (name) {
fout << "\"" << name << "\": {\n";
@ -439,7 +444,7 @@ bool output_json(struct whisper_context * ctx, const char * fname, const whisper
indent++;
};
auto end_obj = [&](bool end = false) {
auto end_obj = [&](bool end) {
indent--;
doindent();
fout << (end ? "}\n" : "},\n");
@ -450,24 +455,24 @@ bool output_json(struct whisper_context * ctx, const char * fname, const whisper
fout << "\"" << name << "\": ";
};
auto value_s = [&](const char *name, const char *val, bool end = false) {
auto value_s = [&](const char *name, const char *val, bool end) {
start_value(name);
char * val_escaped = escape_double_quotes_and_backslashes(val);
fout << "\"" << val_escaped << (end ? "\"\n" : "\",\n");
free(val_escaped);
};
auto end_value = [&](bool end = false) {
auto end_value = [&](bool end) {
fout << (end ? "\n" : ",\n");
};
auto value_i = [&](const char *name, const int64_t val, bool end = false) {
auto value_i = [&](const char *name, const int64_t val, bool end) {
start_value(name);
fout << val;
end_value(end);
};
auto value_b = [&](const char *name, const bool val, bool end = false) {
auto value_b = [&](const char *name, const bool val, bool end) {
start_value(name);
fout << (val ? "true" : "false");
end_value(end);
@ -479,35 +484,35 @@ bool output_json(struct whisper_context * ctx, const char * fname, const whisper
}
fprintf(stderr, "%s: saving output to '%s'\n", __func__, fname);
start_obj();
value_s("systeminfo", whisper_print_system_info());
start_obj(nullptr);
value_s("systeminfo", whisper_print_system_info(), false);
start_obj("model");
value_s("type", whisper_model_type_readable(ctx));
value_b("multilingual", whisper_is_multilingual(ctx));
value_i("vocab", whisper_model_n_vocab(ctx));
value_s("type", whisper_model_type_readable(ctx), false);
value_b("multilingual", whisper_is_multilingual(ctx), false);
value_i("vocab", whisper_model_n_vocab(ctx), false);
start_obj("audio");
value_i("ctx", whisper_model_n_audio_ctx(ctx));
value_i("state", whisper_model_n_audio_state(ctx));
value_i("head", whisper_model_n_audio_head(ctx));
value_i("ctx", whisper_model_n_audio_ctx(ctx), false);
value_i("state", whisper_model_n_audio_state(ctx), false);
value_i("head", whisper_model_n_audio_head(ctx), false);
value_i("layer", whisper_model_n_audio_layer(ctx), true);
end_obj();
end_obj(false);
start_obj("text");
value_i("ctx", whisper_model_n_text_ctx(ctx));
value_i("state", whisper_model_n_text_state(ctx));
value_i("head", whisper_model_n_text_head(ctx));
value_i("ctx", whisper_model_n_text_ctx(ctx), false);
value_i("state", whisper_model_n_text_state(ctx), false);
value_i("head", whisper_model_n_text_head(ctx), false);
value_i("layer", whisper_model_n_text_layer(ctx), true);
end_obj();
value_i("mels", whisper_model_n_mels(ctx));
end_obj(false);
value_i("mels", whisper_model_n_mels(ctx), false);
value_i("ftype", whisper_model_ftype(ctx), true);
end_obj();
end_obj(false);
start_obj("params");
value_s("model", params.model.c_str());
value_s("language", params.language.c_str());
value_s("model", params.model.c_str(), false);
value_s("language", params.language.c_str(), false);
value_b("translate", params.translate, true);
end_obj();
end_obj(false);
start_obj("result");
value_s("language", whisper_lang_str(whisper_full_lang_id(ctx)), true);
end_obj();
end_obj(false);
start_arr("transcription");
const int n_segments = whisper_full_n_segments(ctx);
@ -516,15 +521,15 @@ bool output_json(struct whisper_context * ctx, const char * fname, const whisper
const int64_t t0 = whisper_full_get_segment_t0(ctx, i);
const int64_t t1 = whisper_full_get_segment_t1(ctx, i);
start_obj();
start_obj(nullptr);
start_obj("timestamps");
value_s("from", to_timestamp(t0, true).c_str());
value_s("from", to_timestamp(t0, true).c_str(), false);
value_s("to", to_timestamp(t1, true).c_str(), true);
end_obj();
end_obj(false);
start_obj("offsets");
value_i("from", t0 * 10);
value_i("from", t0 * 10, false);
value_i("to", t1 * 10, true);
end_obj();
end_obj(false);
value_s("text", text, true);
end_obj(i == (n_segments - 1));
}