common.cpp

// -*- mode:c++;indent-tabs-mode:nil;c-basic-offset:4;tab-width:8;coding:utf-8 -*-
// vi: set et ft=cpp ts=4 sts=4 sw=4 fenc=utf-8 :vi
#define _USE_MATH_DEFINES // for M_PI

#include "llamafile/log.h"
#include "llamafile/llamafile.h"
#include "common.h"

// third-party utilities
// use your favorite implementations
// #define DR_WAV_IMPLEMENTATION // [jart] comment out
#include "dr_wav.h"

#if defined(_MSC_VER)
#pragma warning(disable: 4244 4267) // possible loss of data
#endif

#ifdef _WIN32
#include <fcntl.h>
#include <io.h>
#endif

#include <cosmo.h>
#include <stdlib.h>
#include <unistd.h>

#include "third_party/stb/stb_vorbis.h"
#include "miniaudio.h"

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) {
        tinylogf("%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()));
}

bool sam_params_parse(int argc, char ** argv, sam_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 == "-m" || arg == "--model") {
            params.model = argv[++i];
        } else if (arg == "-i" || arg == "--inp") {
            params.fname_inp = argv[++i];
        } else if (arg == "-o" || arg == "--out") {
            params.fname_out = argv[++i];
        } else if (arg == "-h" || arg == "--help") {
            sam_print_usage(argc, argv, params);
            exit(0);
        } else {
            fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
            // sam_print_usage(argc, argv, params); // [jart]
            exit(0);
        }
    }

    return true;
}

void sam_print_usage(int /*argc*/, char ** argv, const sam_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, "  -m FNAME, --model FNAME\n");
    fprintf(stderr, "                        model path (default: %s)\n", params.model.c_str());
    fprintf(stderr, "  -i FNAME, --inp FNAME\n");
    fprintf(stderr, "                        input file (default: %s)\n", params.fname_inp.c_str());
    fprintf(stderr, "  -o FNAME, --out FNAME\n");
    fprintf(stderr, "                        output file (default: %s)\n", params.fname_out.c_str());
    fprintf(stderr, "\n");
}

//  500 -> 00:05.000
// 6000 -> 01:00.000
std::string to_timestamp(int64_t t, bool comma) {
    int64_t msec = t * 10;
    int64_t hr = msec / (1000 * 60 * 60);
    msec = msec - hr * (1000 * 60 * 60);
    int64_t min = msec / (1000 * 60);
    msec = msec - min * (1000 * 60);
    int64_t sec = msec / 1000;
    msec = msec - sec * 1000;

    char buf[32];
    snprintf(buf, sizeof(buf), "%02d:%02d:%02d%s%03d", (int) hr, (int) min, (int) sec, comma ? "," : ".", (int) msec);

    return std::string(buf);
}

int timestamp_to_sample(int64_t t, int n_samples, int whisper_sample_rate) {
    return std::max(0, std::min((int) n_samples - 1, (int) ((t*whisper_sample_rate)/100)));
}

bool is_file_exist(const char *fileName)
{
    std::ifstream infile(fileName);
    return infile.good();
}

bool speak_with_file(const std::string & command, const std::string & text, const std::string & path, int voice_id)
{
    std::ofstream speak_file(path.c_str());
    if (speak_file.fail()) {
        fprintf(stderr, "%s: failed to open speak_file\n", __func__);
        return false;
    } else {
        speak_file.write(text.c_str(), text.size());
        speak_file.close();
        int ret = system((command + " " + std::to_string(voice_id) + " " + path).c_str());
        if (ret != 0) {
            fprintf(stderr, "%s: failed to speak\n", __func__);
            return false;
        }
    }
    return true;
}