mirror of
https://github.com/ggerganov/whisper.cpp.git
synced 2024-11-07 08:34:37 +01:00
506 lines
15 KiB
C++
506 lines
15 KiB
C++
#include "common.h"
|
|
|
|
// third-party utilities
|
|
// use your favorite implementations
|
|
#define DR_WAV_IMPLEMENTATION
|
|
#include "dr_wav.h"
|
|
|
|
#include <cmath>
|
|
#include <fstream>
|
|
#include <regex>
|
|
|
|
#ifndef M_PI
|
|
#define M_PI 3.14159265358979323846
|
|
#endif
|
|
|
|
bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
|
|
for (int i = 1; i < argc; i++) {
|
|
std::string arg = argv[i];
|
|
|
|
if (arg == "-s" || arg == "--seed") {
|
|
params.seed = std::stoi(argv[++i]);
|
|
} else if (arg == "-t" || arg == "--threads") {
|
|
params.n_threads = std::stoi(argv[++i]);
|
|
} else if (arg == "-p" || arg == "--prompt") {
|
|
params.prompt = argv[++i];
|
|
} else if (arg == "-n" || arg == "--n_predict") {
|
|
params.n_predict = std::stoi(argv[++i]);
|
|
} else if (arg == "--top_k") {
|
|
params.top_k = std::stoi(argv[++i]);
|
|
} else if (arg == "--top_p") {
|
|
params.top_p = std::stof(argv[++i]);
|
|
} else if (arg == "--temp") {
|
|
params.temp = std::stof(argv[++i]);
|
|
} else if (arg == "-b" || arg == "--batch_size") {
|
|
params.n_batch = std::stoi(argv[++i]);
|
|
} else if (arg == "-m" || arg == "--model") {
|
|
params.model = argv[++i];
|
|
} else if (arg == "-h" || arg == "--help") {
|
|
gpt_print_usage(argc, argv, params);
|
|
exit(0);
|
|
} else {
|
|
fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
|
|
gpt_print_usage(argc, argv, params);
|
|
exit(0);
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
|
|
fprintf(stderr, "usage: %s [options]\n", argv[0]);
|
|
fprintf(stderr, "\n");
|
|
fprintf(stderr, "options:\n");
|
|
fprintf(stderr, " -h, --help show this help message and exit\n");
|
|
fprintf(stderr, " -s SEED, --seed SEED RNG seed (default: -1)\n");
|
|
fprintf(stderr, " -t N, --threads N number of threads to use during computation (default: %d)\n", params.n_threads);
|
|
fprintf(stderr, " -p PROMPT, --prompt PROMPT\n");
|
|
fprintf(stderr, " prompt to start generation with (default: random)\n");
|
|
fprintf(stderr, " -n N, --n_predict N number of tokens to predict (default: %d)\n", params.n_predict);
|
|
fprintf(stderr, " --top_k N top-k sampling (default: %d)\n", params.top_k);
|
|
fprintf(stderr, " --top_p N top-p sampling (default: %.1f)\n", params.top_p);
|
|
fprintf(stderr, " --temp N temperature (default: %.1f)\n", params.temp);
|
|
fprintf(stderr, " -b N, --batch_size N batch size for prompt processing (default: %d)\n", params.n_batch);
|
|
fprintf(stderr, " -m FNAME, --model FNAME\n");
|
|
fprintf(stderr, " model path (default: %s)\n", params.model.c_str());
|
|
fprintf(stderr, "\n");
|
|
}
|
|
|
|
std::string gpt_random_prompt(std::mt19937 & rng) {
|
|
const int r = rng() % 10;
|
|
switch (r) {
|
|
case 0: return "So";
|
|
case 1: return "Once upon a time";
|
|
case 2: return "When";
|
|
case 3: return "The";
|
|
case 4: return "After";
|
|
case 5: return "If";
|
|
case 6: return "import";
|
|
case 7: return "He";
|
|
case 8: return "She";
|
|
case 9: return "They";
|
|
default: return "To";
|
|
}
|
|
|
|
return "The";
|
|
}
|
|
|
|
std::string trim(const std::string & s) {
|
|
std::regex e("^\\s+|\\s+$");
|
|
return std::regex_replace(s, e, "");
|
|
}
|
|
|
|
std::string replace(const std::string & s, const std::string & from, const std::string & to) {
|
|
std::string result = s;
|
|
size_t pos = 0;
|
|
while ((pos = result.find(from, pos)) != std::string::npos) {
|
|
result.replace(pos, from.length(), to);
|
|
pos += to.length();
|
|
}
|
|
return result;
|
|
}
|
|
|
|
std::map<std::string, int32_t> json_parse(const std::string & fname) {
|
|
std::map<std::string, int32_t> result;
|
|
|
|
// read file into string
|
|
std::string json;
|
|
{
|
|
std::ifstream ifs(fname);
|
|
if (!ifs) {
|
|
fprintf(stderr, "Failed to open %s\n", fname.c_str());
|
|
exit(1);
|
|
}
|
|
|
|
json = std::string((std::istreambuf_iterator<char>(ifs)),
|
|
(std::istreambuf_iterator<char>()));
|
|
}
|
|
|
|
if (json[0] != '{') {
|
|
return result;
|
|
}
|
|
|
|
// parse json
|
|
{
|
|
bool has_key = false;
|
|
bool in_token = false;
|
|
|
|
std::string str_key = "";
|
|
std::string str_val = "";
|
|
|
|
int n = json.size();
|
|
for (int i = 1; i < n; ++i) {
|
|
if (!in_token) {
|
|
if (json[i] == ' ') continue;
|
|
if (json[i] == '"') {
|
|
in_token = true;
|
|
continue;
|
|
}
|
|
} else {
|
|
if (json[i] == '\\' && i+1 < n) {
|
|
if (has_key == false) {
|
|
str_key += json[i];
|
|
} else {
|
|
str_val += json[i];
|
|
}
|
|
++i;
|
|
} else if (json[i] == '"') {
|
|
if (has_key == false) {
|
|
has_key = true;
|
|
++i;
|
|
while (json[i] == ' ') ++i;
|
|
++i; // :
|
|
while (json[i] == ' ') ++i;
|
|
if (json[i] != '\"') {
|
|
while (json[i] != ',' && json[i] != '}') {
|
|
str_val += json[i++];
|
|
}
|
|
has_key = false;
|
|
} else {
|
|
in_token = true;
|
|
continue;
|
|
}
|
|
} else {
|
|
has_key = false;
|
|
}
|
|
|
|
str_key = ::replace(str_key, "\\u0120", " " ); // \u0120 -> space
|
|
str_key = ::replace(str_key, "\\u010a", "\n"); // \u010a -> new line
|
|
str_key = ::replace(str_key, "\\\"", "\""); // \\\" -> "
|
|
|
|
try {
|
|
result[str_key] = std::stoi(str_val);
|
|
} catch (...) {
|
|
//fprintf(stderr, "%s: ignoring key '%s' with value '%s'\n", fname.c_str(), str_key.c_str(), str_val.c_str());
|
|
|
|
}
|
|
str_key = "";
|
|
str_val = "";
|
|
in_token = false;
|
|
continue;
|
|
}
|
|
if (has_key == false) {
|
|
str_key += json[i];
|
|
} else {
|
|
str_val += json[i];
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
return result;
|
|
}
|
|
|
|
std::vector<gpt_vocab::id> gpt_tokenize(const gpt_vocab & vocab, const std::string & text) {
|
|
std::vector<std::string> words;
|
|
|
|
// first split the text into words
|
|
{
|
|
std::string str = text;
|
|
std::string pat = R"('s|'t|'re|'ve|'m|'ll|'d| ?[[:alpha:]]+| ?[[:digit:]]+| ?[^\s[:alpha:][:digit:]]+|\s+(?!\S)|\s+)";
|
|
|
|
std::regex re(pat);
|
|
std::smatch m;
|
|
|
|
while (std::regex_search(str, m, re)) {
|
|
for (auto x : m) {
|
|
words.push_back(x);
|
|
}
|
|
str = m.suffix();
|
|
}
|
|
}
|
|
|
|
// find the longest tokens that form the words:
|
|
std::vector<gpt_vocab::id> tokens;
|
|
for (const auto & word : words) {
|
|
if (word.size() == 0) continue;
|
|
|
|
int i = 0;
|
|
int n = word.size();
|
|
while (i < n) {
|
|
int j = n;
|
|
while (j > i) {
|
|
auto it = vocab.token_to_id.find(word.substr(i, j-i));
|
|
if (it != vocab.token_to_id.end()) {
|
|
tokens.push_back(it->second);
|
|
i = j;
|
|
break;
|
|
}
|
|
--j;
|
|
}
|
|
if (i == n) {
|
|
break;
|
|
}
|
|
if (j == i) {
|
|
auto sub = word.substr(i, 1);
|
|
if (vocab.token_to_id.find(sub) != vocab.token_to_id.end()) {
|
|
tokens.push_back(vocab.token_to_id.at(sub));
|
|
} else {
|
|
fprintf(stderr, "%s: unknown token '%s'\n", __func__, sub.data());
|
|
}
|
|
++i;
|
|
}
|
|
}
|
|
}
|
|
|
|
return tokens;
|
|
}
|
|
|
|
bool gpt_vocab_init(const std::string & fname, gpt_vocab & vocab) {
|
|
printf("%s: loading vocab from '%s'\n", __func__, fname.c_str());
|
|
|
|
vocab.token_to_id = ::json_parse(fname);
|
|
|
|
for (const auto & kv : vocab.token_to_id) {
|
|
vocab.id_to_token[kv.second] = kv.first;
|
|
}
|
|
|
|
printf("%s: vocab size = %d\n", __func__, (int) vocab.token_to_id.size());
|
|
|
|
// print the vocabulary
|
|
//for (auto kv : vocab.token_to_id) {
|
|
// printf("'%s' -> %d\n", kv.first.data(), kv.second);
|
|
//}
|
|
|
|
return true;
|
|
}
|
|
|
|
gpt_vocab::id gpt_sample_top_k_top_p(
|
|
const gpt_vocab & vocab,
|
|
const float * logits,
|
|
int top_k,
|
|
double top_p,
|
|
double temp,
|
|
std::mt19937 & rng) {
|
|
int n_logits = vocab.id_to_token.size();
|
|
|
|
std::vector<std::pair<double, gpt_vocab::id>> logits_id;
|
|
logits_id.reserve(n_logits);
|
|
|
|
{
|
|
const double scale = 1.0/temp;
|
|
for (int i = 0; i < n_logits; ++i) {
|
|
logits_id.push_back(std::make_pair(logits[i]*scale, i));
|
|
}
|
|
}
|
|
|
|
// find the top K tokens
|
|
std::partial_sort(
|
|
logits_id.begin(),
|
|
logits_id.begin() + top_k, logits_id.end(),
|
|
[](const std::pair<double, gpt_vocab::id> & a, const std::pair<double, gpt_vocab::id> & b) {
|
|
return a.first > b.first;
|
|
});
|
|
|
|
logits_id.resize(top_k);
|
|
|
|
double maxl = -INFINITY;
|
|
for (const auto & kv : logits_id) {
|
|
maxl = std::max(maxl, kv.first);
|
|
}
|
|
|
|
// compute probs for the top K tokens
|
|
std::vector<double> probs;
|
|
probs.reserve(logits_id.size());
|
|
|
|
double sum = 0.0;
|
|
for (const auto & kv : logits_id) {
|
|
double p = exp(kv.first - maxl);
|
|
probs.push_back(p);
|
|
sum += p;
|
|
}
|
|
|
|
// normalize the probs
|
|
for (auto & p : probs) {
|
|
p /= sum;
|
|
}
|
|
|
|
if (top_p < 1.0f) {
|
|
double cumsum = 0.0f;
|
|
for (int i = 0; i < top_k; i++) {
|
|
cumsum += probs[i];
|
|
if (cumsum >= top_p) {
|
|
top_k = i + 1;
|
|
probs.resize(top_k);
|
|
logits_id.resize(top_k);
|
|
break;
|
|
}
|
|
}
|
|
|
|
cumsum = 1.0/cumsum;
|
|
for (int i = 0; i < (int) probs.size(); i++) {
|
|
probs[i] *= cumsum;
|
|
}
|
|
}
|
|
|
|
//printf("\n");
|
|
//for (int i = 0; i < (int) probs.size(); i++) {
|
|
// printf("%d: '%s' %f\n", i, vocab.id_to_token.at(logits_id[i].second).c_str(), probs[i]);
|
|
//}
|
|
//exit(0);
|
|
|
|
std::discrete_distribution<> dist(probs.begin(), probs.end());
|
|
int idx = dist(rng);
|
|
|
|
return logits_id[idx].second;
|
|
}
|
|
|
|
bool read_wav(const std::string & fname, std::vector<float>& pcmf32, std::vector<std::vector<float>>& pcmf32s, bool stereo) {
|
|
drwav wav;
|
|
std::vector<uint8_t> wav_data; // used for pipe input from stdin
|
|
|
|
if (fname == "-") {
|
|
{
|
|
uint8_t buf[1024];
|
|
while (true)
|
|
{
|
|
const size_t n = fread(buf, 1, sizeof(buf), stdin);
|
|
if (n == 0) {
|
|
break;
|
|
}
|
|
wav_data.insert(wav_data.end(), buf, buf + n);
|
|
}
|
|
}
|
|
|
|
if (drwav_init_memory(&wav, wav_data.data(), wav_data.size(), nullptr) == false) {
|
|
fprintf(stderr, "error: failed to open WAV file from stdin\n");
|
|
return false;
|
|
}
|
|
|
|
fprintf(stderr, "%s: read %zu bytes from stdin\n", __func__, wav_data.size());
|
|
}
|
|
else if (drwav_init_file(&wav, fname.c_str(), nullptr) == false) {
|
|
fprintf(stderr, "error: failed to open '%s' as WAV file\n", fname.c_str());
|
|
return false;
|
|
}
|
|
|
|
if (wav.channels != 1 && wav.channels != 2) {
|
|
fprintf(stderr, "%s: WAV file '%s' must be mono or stereo\n", __func__, fname.c_str());
|
|
return false;
|
|
}
|
|
|
|
if (stereo && wav.channels != 2) {
|
|
fprintf(stderr, "%s: WAV file '%s' must be stereo for diarization\n", __func__, fname.c_str());
|
|
return false;
|
|
}
|
|
|
|
if (wav.sampleRate != COMMON_SAMPLE_RATE) {
|
|
fprintf(stderr, "%s: WAV file '%s' must be %i kHz\n", __func__, fname.c_str(), COMMON_SAMPLE_RATE/1000);
|
|
return false;
|
|
}
|
|
|
|
if (wav.bitsPerSample != 16) {
|
|
fprintf(stderr, "%s: WAV file '%s' must be 16-bit\n", __func__, fname.c_str());
|
|
return false;
|
|
}
|
|
|
|
const uint64_t n = wav_data.empty() ? wav.totalPCMFrameCount : wav_data.size()/(wav.channels*wav.bitsPerSample/8);
|
|
|
|
std::vector<int16_t> pcm16;
|
|
pcm16.resize(n*wav.channels);
|
|
drwav_read_pcm_frames_s16(&wav, n, pcm16.data());
|
|
drwav_uninit(&wav);
|
|
|
|
// convert to mono, float
|
|
pcmf32.resize(n);
|
|
if (wav.channels == 1) {
|
|
for (uint64_t i = 0; i < n; i++) {
|
|
pcmf32[i] = float(pcm16[i])/32768.0f;
|
|
}
|
|
} else {
|
|
for (uint64_t i = 0; i < n; i++) {
|
|
pcmf32[i] = float(pcm16[2*i] + pcm16[2*i + 1])/65536.0f;
|
|
}
|
|
}
|
|
|
|
if (stereo) {
|
|
// convert to stereo, float
|
|
pcmf32s.resize(2);
|
|
|
|
pcmf32s[0].resize(n);
|
|
pcmf32s[1].resize(n);
|
|
for (uint64_t i = 0; i < n; i++) {
|
|
pcmf32s[0][i] = float(pcm16[2*i])/32768.0f;
|
|
pcmf32s[1][i] = float(pcm16[2*i + 1])/32768.0f;
|
|
}
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
void high_pass_filter(std::vector<float> & data, float cutoff, float sample_rate) {
|
|
const float rc = 1.0f / (2.0f * M_PI * cutoff);
|
|
const float dt = 1.0f / sample_rate;
|
|
const float alpha = dt / (rc + dt);
|
|
|
|
float y = data[0];
|
|
|
|
for (size_t i = 1; i < data.size(); i++) {
|
|
y = alpha * (y + data[i] - data[i - 1]);
|
|
data[i] = y;
|
|
}
|
|
}
|
|
|
|
bool vad_simple(std::vector<float> & pcmf32, int sample_rate, int last_ms, float vad_thold, float freq_thold, bool verbose) {
|
|
const int n_samples = pcmf32.size();
|
|
const int n_samples_last = (sample_rate * last_ms) / 1000;
|
|
|
|
if (n_samples_last >= n_samples) {
|
|
// not enough samples - assume no speech
|
|
return false;
|
|
}
|
|
|
|
if (freq_thold > 0.0f) {
|
|
high_pass_filter(pcmf32, freq_thold, sample_rate);
|
|
}
|
|
|
|
float energy_all = 0.0f;
|
|
float energy_last = 0.0f;
|
|
|
|
for (int i = 0; i < n_samples; i++) {
|
|
energy_all += fabsf(pcmf32[i]);
|
|
|
|
if (i >= n_samples - n_samples_last) {
|
|
energy_last += fabsf(pcmf32[i]);
|
|
}
|
|
}
|
|
|
|
energy_all /= n_samples;
|
|
energy_last /= n_samples_last;
|
|
|
|
if (verbose) {
|
|
fprintf(stderr, "%s: energy_all: %f, energy_last: %f, vad_thold: %f, freq_thold: %f\n", __func__, energy_all, energy_last, vad_thold, freq_thold);
|
|
}
|
|
|
|
if (energy_last > vad_thold*energy_all) {
|
|
return false;
|
|
}
|
|
|
|
return true;
|
|
}
|
|
|
|
float similarity(const std::string & s0, const std::string & s1) {
|
|
const size_t len0 = s0.size() + 1;
|
|
const size_t len1 = s1.size() + 1;
|
|
|
|
std::vector<int> col(len1, 0);
|
|
std::vector<int> prevCol(len1, 0);
|
|
|
|
for (size_t i = 0; i < len1; i++) {
|
|
prevCol[i] = i;
|
|
}
|
|
|
|
for (size_t i = 0; i < len0; i++) {
|
|
col[0] = i;
|
|
for (size_t j = 1; j < len1; j++) {
|
|
col[j] = std::min(std::min(1 + col[j - 1], 1 + prevCol[j]), prevCol[j - 1] + (i > 0 && s0[i - 1] == s1[j - 1] ? 0 : 1));
|
|
}
|
|
col.swap(prevCol);
|
|
}
|
|
|
|
const float dist = prevCol[len1 - 1];
|
|
|
|
return 1.0f - (dist / std::max(s0.size(), s1.size()));
|
|
}
|