public_git/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggml-org/whisper.cpp.git synced 2026-10-07 15:01:16 +02:00
Code Issues Packages Projects Releases Wiki Activity

whisper : remove speed_up and phase_vocoder* functions (#2198)

Browse Source 
* whisper : fix cast warning

* whisper : remove phase_vocoder functions, ref #2195

* whisper : remove speed_up from whisper_full_params, closes #2195
This commit is contained in:
Borislav Stanimirov
2024-05-31 11:37:29 +03:00
committed by GitHub
parent b87494bb8f
commit af5833e298
20 changed files with 14 additions and 161 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/addon.node/addon.cpp
-3
View File
@@ -25,7 +25,6 @@ struct whisper_params {
float entropy_thold = 2.4f;
float logprob_thold = -1.0f;
bool speed_up = false;
bool translate = false;
bool diarize = false;
bool output_txt = false;
@@ -232,8 +231,6 @@ int run(whisper_params &params, std::vector<std::vector<std::string>> &result) {
wparams.max_len = params.output_wts && params.max_len == 0 ? 60 : params.max_len;
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
wparams.greedy.best_of = params.best_of;
wparams.beam_search.beam_size = params.beam_size;
examples/command/command.cpp
-5
View File
@@ -38,7 +38,6 @@ struct whisper_params {
grammar_parser::parse_state grammar_parsed;
bool speed_up = false;
bool translate = false;
bool print_special = false;
bool print_energy = false;
@@ -76,7 +75,6 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
else if (arg == "-ac" || arg == "--audio-ctx") { params.audio_ctx = std::stoi(argv[++i]); }
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]); }
else if (arg == "-su" || arg == "--speed-up") { params.speed_up = true; }
else if (arg == "-tr" || arg == "--translate") { params.translate = true; }
else if (arg == "-ps" || arg == "--print-special") { params.print_special = true; }
else if (arg == "-pe" || arg == "--print-energy") { params.print_energy = true; }
@@ -115,7 +113,6 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
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, " -ps, --print-special [%-7s] print special tokens\n", params.print_special ? "true" : "false");
fprintf(stderr, " -pe, --print-energy [%-7s] print sound energy (for debugging)\n", params.print_energy ? "true" : "false");
@@ -165,7 +162,6 @@ std::string transcribe(
wparams.n_threads = params.n_threads;
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
wparams.temperature = 0.4f;
wparams.temperature_inc = 1.0f;
@@ -371,7 +367,6 @@ int process_command_list(struct whisper_context * ctx, audio_async &audio, const
wparams.n_threads = params.n_threads;
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
wparams.prompt_tokens = k_tokens.data();
wparams.prompt_n_tokens = k_tokens.size();
examples/common.h
+1 -1
View File
@@ -185,7 +185,7 @@ private:
// It is assumed that PCM data is normalized to a range from -1 to 1
bool write_audio(const float * data, size_t length) {
for (size_t i = 0; i < length; ++i) {
const int16_t intSample = data[i] * 32767;
const int16_t intSample = int16_t(data[i] * 32767);
file.write(reinterpret_cast<const char *>(&intSample), sizeof(int16_t));
dataSize += sizeof(int16_t);
}
examples/lsp/lsp.cpp
-5
View File
@@ -26,7 +26,6 @@ struct whisper_params {
float vad_thold = 0.6f;
float freq_thold = 100.0f;
bool speed_up = false;
bool translate = false;
bool print_special = false;
bool print_energy = false;
@@ -70,7 +69,6 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
else if (arg == "-ac" || arg == "--audio-ctx") { params.audio_ctx = std::stoi(argv[++i]); }
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]); }
else if (arg == "-su" || arg == "--speed-up") { params.speed_up = true; }
else if (arg == "-tr" || arg == "--translate") { params.translate = true; }
else if (arg == "-ps" || arg == "--print-special") { params.print_special = true; }
else if (arg == "-pe" || arg == "--print-energy") { params.print_energy = true; }
@@ -102,7 +100,6 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
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, " -ps, --print-special [%-7s] print special tokens\n", params.print_special ? "true" : "false");
fprintf(stderr, " -pe, --print-energy [%-7s] print sound energy (for debugging)\n", params.print_energy ? "true" : "false");
@@ -184,7 +181,6 @@ json unguided_transcription(struct whisper_context * ctx, audio_async &audio, js
wparams.n_threads = params.n_threads;
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
wparams.suppress_non_speech_tokens = true;
// run the transformer and a single decoding pass
if (whisper_full(ctx, wparams, pcmf32.data(), pcmf32.size()) != 0) {
@@ -223,7 +219,6 @@ json guided_transcription(struct whisper_context * ctx, audio_async &audio, cons
wparams.n_threads = params.n_threads;
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
// TODO: Do some time testing. Does an overly long prompt slow down processing?
// Set up command sets/precompute prompts
examples/main/main.cpp
-4
View File
@@ -47,7 +47,6 @@ struct whisper_params {
float temperature = 0.0f;
float temperature_inc = 0.2f;
bool speed_up = false;
bool debug_mode = false;
bool translate = false;
bool detect_language = false;
@@ -138,7 +137,6 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
else if (arg == "-lpt" || arg == "--logprob-thold") { params.logprob_thold = std::stof(argv[++i]); }
else if (arg == "-tp" || arg == "--temperature") { params.temperature = std::stof(argv[++i]); }
else if (arg == "-tpi" || arg == "--temperature-inc") { params.temperature_inc = std::stof(argv[++i]); }
// else if (arg == "-su" || arg == "--speed-up") { params.speed_up = true; }
else if (arg == "-debug"|| arg == "--debug-mode") { params.debug_mode = true; }
else if (arg == "-tr" || arg == "--translate") { params.translate = true; }
else if (arg == "-di" || arg == "--diarize") { params.diarize = true; }
@@ -206,7 +204,6 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
fprintf(stderr, " -lpt N, --logprob-thold N [%-7.2f] log probability threshold for decoder fail\n", params.logprob_thold);
fprintf(stderr, " -tp, --temperature N [%-7.2f] The sampling temperature, between 0 and 1\n", params.temperature);
fprintf(stderr, " -tpi, --temperature-inc N [%-7.2f] The increment of temperature, between 0 and 1\n",params.temperature_inc);
// fprintf(stderr, " -su, --speed-up [%-7s] speed up audio by x2 (reduced accuracy)\n", params.speed_up ? "true" : "false");
fprintf(stderr, " -debug, --debug-mode [%-7s] enable debug mode (eg. dump log_mel)\n", params.debug_mode ? "true" : "false");
fprintf(stderr, " -tr, --translate [%-7s] translate from source language to english\n", params.translate ? "true" : "false");
fprintf(stderr, " -di, --diarize [%-7s] stereo audio diarization\n", params.diarize ? "true" : "false");
@@ -1106,7 +1103,6 @@ int main(int argc, char ** argv) {
wparams.split_on_word = params.split_on_word;
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
wparams.debug_mode = params.debug_mode;
wparams.tdrz_enable = params.tinydiarize; // [TDRZ]
examples/server/server.cpp
-4
View File
@@ -61,7 +61,6 @@ struct whisper_params {
float temperature = 0.00f;
float temperature_inc = 0.20f;
bool speed_up = false;
bool debug_mode = false;
bool translate = false;
bool detect_language = false;
@@ -112,7 +111,6 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
fprintf(stderr, " -wt N, --word-thold N [%-7.2f] word timestamp probability threshold\n", params.word_thold);
fprintf(stderr, " -et N, --entropy-thold N [%-7.2f] entropy threshold for decoder fail\n", params.entropy_thold);
fprintf(stderr, " -lpt N, --logprob-thold N [%-7.2f] log probability threshold for decoder fail\n", params.logprob_thold);
// fprintf(stderr, " -su, --speed-up [%-7s] speed up audio by x2 (reduced accuracy)\n", params.speed_up ? "true" : "false");
fprintf(stderr, " -debug, --debug-mode [%-7s] enable debug mode (eg. dump log_mel)\n", params.debug_mode ? "true" : "false");
fprintf(stderr, " -tr, --translate [%-7s] translate from source language to english\n", params.translate ? "true" : "false");
fprintf(stderr, " -di, --diarize [%-7s] stereo audio diarization\n", params.diarize ? "true" : "false");
@@ -159,7 +157,6 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params, serve
else if (arg == "-wt" || arg == "--word-thold") { params.word_thold = std::stof(argv[++i]); }
else if (arg == "-et" || arg == "--entropy-thold") { params.entropy_thold = std::stof(argv[++i]); }
else if (arg == "-lpt" || arg == "--logprob-thold") { params.logprob_thold = std::stof(argv[++i]); }
// else if (arg == "-su" || arg == "--speed-up") { params.speed_up = true; }
else if (arg == "-debug"|| arg == "--debug-mode") { params.debug_mode = true; }
else if (arg == "-tr" || arg == "--translate") { params.translate = true; }
else if (arg == "-di" || arg == "--diarize") { params.diarize = true; }
@@ -768,7 +765,6 @@ int main(int argc, char ** argv) {
wparams.split_on_word = params.split_on_word;
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
wparams.debug_mode = params.debug_mode;
wparams.tdrz_enable = params.tinydiarize; // [TDRZ]
examples/stream/stream.cpp
-4
View File
@@ -27,7 +27,6 @@ struct whisper_params {
float vad_thold = 0.6f;
float freq_thold = 100.0f;
bool speed_up = false;
bool translate = false;
bool no_fallback = false;
bool print_special = false;
@@ -62,7 +61,6 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
else if (arg == "-ac" || arg == "--audio-ctx") { params.audio_ctx = std::stoi(argv[++i]); }
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]); }
else if (arg == "-su" || arg == "--speed-up") { params.speed_up = true; }
else if (arg == "-tr" || arg == "--translate") { params.translate = true; }
else if (arg == "-nf" || arg == "--no-fallback") { params.no_fallback = true; }
else if (arg == "-ps" || arg == "--print-special") { params.print_special = true; }
@@ -100,7 +98,6 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
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");
@@ -314,7 +311,6 @@ int main(int argc, char ** argv) {
wparams.n_threads = params.n_threads;
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
wparams.tdrz_enable = params.tinydiarize; // [TDRZ]
examples/talk-llama/talk-llama.cpp
-4
View File
@@ -59,7 +59,6 @@ struct whisper_params {
float vad_thold = 0.6f;
float freq_thold = 100.0f;
bool speed_up = false;
bool translate = false;
bool print_special = false;
bool print_energy = false;
@@ -100,7 +99,6 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
else if (arg == "-ngl" || arg == "--n-gpu-layers") { params.n_gpu_layers = std::stoi(argv[++i]); }
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]); }
else if (arg == "-su" || arg == "--speed-up") { params.speed_up = true; }
else if (arg == "-tr" || arg == "--translate") { params.translate = true; }
else if (arg == "-ps" || arg == "--print-special") { params.print_special = true; }
else if (arg == "-pe" || arg == "--print-energy") { params.print_energy = true; }
@@ -149,7 +147,6 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
fprintf(stderr, " -ngl N, --n-gpu-layers N [%-7d] number of layers to store in VRAM\n", params.n_gpu_layers);
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, " -ps, --print-special [%-7s] print special tokens\n", params.print_special ? "true" : "false");
fprintf(stderr, " -pe, --print-energy [%-7s] print sound energy (for debugging)\n", params.print_energy ? "true" : "false");
@@ -205,7 +202,6 @@ std::string transcribe(
wparams.prompt_n_tokens = prompt_tokens.empty() ? 0 : prompt_tokens.size();
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
if (whisper_full(ctx, wparams, pcmf32.data(), pcmf32.size()) != 0) {
return "";
examples/talk/talk.cpp
-4
View File
@@ -26,7 +26,6 @@ struct whisper_params {
float vad_thold = 0.6f;
float freq_thold = 100.0f;
bool speed_up = false;
bool translate = false;
bool print_special = false;
bool print_energy = false;
@@ -60,7 +59,6 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
else if (arg == "-ac" || arg == "--audio-ctx") { params.audio_ctx = std::stoi(argv[++i]); }
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]); }
else if (arg == "-su" || arg == "--speed-up") { params.speed_up = true; }
else if (arg == "-tr" || arg == "--translate") { params.translate = true; }
else if (arg == "-ps" || arg == "--print-special") { params.print_special = true; }
else if (arg == "-pe" || arg == "--print-energy") { params.print_energy = true; }
@@ -96,7 +94,6 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
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, " -ps, --print-special [%-7s] print special tokens\n", params.print_special ? "true" : "false");
fprintf(stderr, " -pe, --print-energy [%-7s] print sound energy (for debugging)\n", params.print_energy ? "true" : "false");
@@ -132,7 +129,6 @@ std::string transcribe(whisper_context * ctx, const whisper_params & params, con
wparams.n_threads = params.n_threads;
wparams.audio_ctx = params.audio_ctx;
wparams.speed_up = params.speed_up;
if (whisper_full(ctx, wparams, pcmf32.data(), pcmf32.size()) != 0) {
return "";
examples/wchess/wchess.cmd/wchess.cmd.cpp
-3
View File
@@ -26,7 +26,6 @@ struct whisper_params {
float grammar_penalty = 100.0f;
bool speed_up = false;
bool translate = false;
bool print_special = false;
bool print_energy = false;
@@ -57,7 +56,6 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
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, " -ps, --print-special [%-7s] print special tokens\n", params.print_special ? "true" : "false");
fprintf(stderr, " -pe, --print-energy [%-7s] print sound energy (for debugging)\n", params.print_energy ? "true" : "false");
@@ -89,7 +87,6 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
else if (arg == "-ac" || arg == "--audio-ctx") { params.audio_ctx = std::stoi(argv[++i]); }
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]); }
else if (arg == "-su" || arg == "--speed-up") { params.speed_up = true; }
else if (arg == "-tr" || arg == "--translate") { params.translate = true; }
else if (arg == "-ps" || arg == "--print-special") { params.print_special = true; }
else if (arg == "-pe" || arg == "--print-energy") { params.print_energy = true; }