whispercpp 1.3.1 → 1.3.2

data/ext/sources/examples/wchess/libwchess/Chessboard.h

@@ -0,0 +1,33 @@
+#pragma once
+#include <string>
+#include <set>
+#include <memory>
+
+// just basic validation
+// fixme: missing en passant, castling, promotion, etc.
+struct State;
+class Piece;
+class Chessboard {
+public:
+    Chessboard();
+    ~Chessboard();
+    std::string process(const std::string& command);
+    std::string stringifyBoard();
+    const std::string& grammar() { return m_grammar; }
+    const std::string& prompt() { return m_prompt; }
+    void setPrompt(const std::string& prompt);
+private:
+    bool parseCommand(const std::string& command, Piece*& piece, char& pos_to);
+    bool move(Piece& piece, char pos);
+    void flagUpdates(char pos_from, char pos_to);
+    void updatePins(Piece& piece);
+    void detectChecks();
+    void setGrammar();
+
+    std::unique_ptr<State> m_state;
+    std::set<char> m_allowedInCheck;
+    bool m_inCheck = false;
+    int m_moveCounter = 0;
+    std::string m_grammar;
+    std::string m_prompt;
+};

data/ext/sources/examples/wchess/libwchess/WChess.cpp

@@ -0,0 +1,193 @@
+#include "WChess.h"
+#include "Chessboard.h"
+#include "grammar-parser.h"
+#include "common.h"
+#include <chrono>
+
+WChess::WChess(whisper_context * ctx,
+        const whisper_full_params & wparams,
+        callbacks cb,
+        settings s)
+        : m_ctx(ctx)
+        , m_wparams(wparams)
+        , m_cb(cb)
+        , m_settings(s)
+        , m_board(new Chessboard())
+{}
+
+WChess::~WChess() = default;
+
+void WChess::set_move(const std::string& moves, float prob) const {
+    if (m_cb.set_move) (*m_cb.set_move)(moves, prob);
+}
+
+void WChess::set_grammar(const std::string& grammar) const {
+    if (m_cb.set_grammar) (*m_cb.set_grammar)(grammar);
+}
+
+bool WChess::get_audio(std::vector<float>& pcmf32) const {
+    if (m_cb.get_audio) return (*m_cb.get_audio)(pcmf32);
+    return false;
+}
+
+std::string WChess::stringify_board() const {
+    return m_board->stringifyBoard();
+}
+
+std::string WChess::get_grammar() const {
+    return m_board->grammar();
+}
+
+void WChess::run() {
+    bool have_prompt  = true;
+    bool ask_prompt   = !have_prompt;
+
+    float logprob_min  = 0.0f;
+
+    float logprob_sum  = 0.0f;
+
+    int n_tokens  = 0;
+
+    std::vector<float> pcmf32_cur;
+    std::vector<float> pcmf32_prompt;
+
+    const std::string k_prompt = have_prompt ? "" : "rook to d4, f3";
+    int64_t t_ms = 0;
+
+    if (ask_prompt) {
+        fprintf(stdout, "\n");
+        fprintf(stdout, "%s: Say the following phrase: '%s%s%s'\n", __func__, "\033[1m", k_prompt.c_str(), "\033[0m");
+        fprintf(stdout, "\n");
+
+        ask_prompt = false;
+    }
+
+    while (get_audio(pcmf32_cur)) {
+        if (!pcmf32_cur.empty()) {
+            // fprintf(stdout, "%s: Processing ...\n", __func__);
+
+            if (!have_prompt) {
+                const auto txt = ::trim(transcribe(pcmf32_cur, logprob_min, logprob_sum, n_tokens, t_ms));
+
+                fprintf(stdout, "%s: Heard '%s%s%s', (t = %d ms)\n", __func__, "\033[1m", txt.c_str(), "\033[0m", (int) t_ms);
+
+                const float sim = similarity(txt, k_prompt);
+
+                if (txt.length() < 0.8*k_prompt.length() || txt.length() > 1.2*k_prompt.length() || sim < 0.8f) {
+                    fprintf(stdout, "%s: WARNING: prompt not recognized, try again\n", __func__);
+                    ask_prompt = true;
+                } else {
+                    fprintf(stdout, "\n");
+                    fprintf(stdout, "%s: The prompt has been recognized!\n", __func__);
+                    fprintf(stdout, "%s: Waiting for voice commands ...\n", __func__);
+                    fprintf(stdout, "\n");
+
+                    // save the audio for the prompt
+                    pcmf32_prompt = pcmf32_cur;
+                    have_prompt = true;
+                    m_board->setPrompt(k_prompt);
+                }
+            } else {
+                if (!pcmf32_prompt.empty()) pcmf32_cur.insert(pcmf32_cur.begin(), pcmf32_prompt.begin(), pcmf32_prompt.end());
+                constexpr size_t MIN_SIZE = 1.2 * WHISPER_SAMPLE_RATE;
+                if (MIN_SIZE > pcmf32_cur.size()) pcmf32_cur.insert(pcmf32_cur.begin(), MIN_SIZE - pcmf32_cur.size(), 0.0f);
+
+                // fprintf(stdout, "%s: grammar rules:\n'%s'\n", __func__, m_board->grammar().c_str());
+
+                auto grammar_parsed = grammar_parser::parse(m_board->grammar().c_str());
+                auto grammar_rules  = grammar_parsed.c_rules();
+
+                m_wparams.grammar_rules   = grammar_rules.data();
+                m_wparams.n_grammar_rules = grammar_rules.size();
+
+                m_wparams.i_start_rule    = grammar_parsed.symbol_ids.at("move");
+                auto txt = ::trim(transcribe(pcmf32_cur, logprob_min, logprob_sum, n_tokens, t_ms));
+
+                const float p = 100.0f * std::exp(logprob_min);
+
+                fprintf(stdout, "%s: heard '%s'\n", __func__, txt.c_str());
+
+                // find the prompt in the text
+                float best_sim = 0.0f;
+                size_t best_len = 0;
+                for (int n = 0.8*k_prompt.size(); n <= 1.2*k_prompt.size(); ++n) {
+                    const auto prompt = txt.substr(0, n);
+
+                    const float sim = similarity(prompt, k_prompt);
+
+                    //fprintf(stderr, "%s: prompt = '%s', sim = %f\n", __func__, prompt.c_str(), sim);
+
+                    if (sim > best_sim) {
+                        best_sim = sim;
+                        best_len = n;
+                    }
+                }
+
+                fprintf(stdout, "%s:   DEBUG: txt = '%s', prob = %.2f%%\n", __func__, txt.c_str(), p);
+                std::string command = ::trim(txt.substr(best_len));
+
+                fprintf(stdout, "%s: Command '%s%s%s', (t = %d ms)\n", __func__, "\033[1m", command.c_str(), "\033[0m", (int) t_ms);
+                fprintf(stdout, "\n");
+
+                if (!command.empty()) {
+                    set_move(m_board->process(command), p);
+                    set_grammar(m_board->grammar());
+                }
+                if (m_board->grammar().empty()) {
+                    fprintf(stdout, "%s: No more moves possible\n", __func__);
+                    break;
+                }
+            }
+        }
+
+        if (ask_prompt) {
+            fprintf(stdout, "\n");
+            fprintf(stdout, "%s: Say the following phrase: '%s%s%s'\n", __func__, "\033[1m", k_prompt.c_str(), "\033[0m");
+            fprintf(stdout, "\n");
+
+            ask_prompt = false;
+        }
+    }
+}
+
+std::string WChess::transcribe(
+                const std::vector<float> & pcmf32,
+                float & logprob_min,
+                float & logprob_sum,
+                int & n_tokens,
+                int64_t & t_ms) {
+    const auto t_start = std::chrono::high_resolution_clock::now();
+
+    logprob_min = 0.0f;
+    logprob_sum = 0.0f;
+    n_tokens    = 0;
+    t_ms = 0;
+
+    if (whisper_full(m_ctx, m_wparams, pcmf32.data(), pcmf32.size()) != 0) {
+        return {};
+    }
+
+    std::string result;
+
+    const int n_segments = whisper_full_n_segments(m_ctx);
+    for (int i = 0; i < n_segments; ++i) {
+        const char * text = whisper_full_get_segment_text(m_ctx, i);
+
+        result += text;
+
+        const int n = whisper_full_n_tokens(m_ctx, i);
+        for (int j = 0; j < n; ++j) {
+            const auto token = whisper_full_get_token_data(m_ctx, i, j);
+
+            if(token.plog > 0.0f) return {};
+            logprob_min = std::min(logprob_min, token.plog);
+            logprob_sum += token.plog;
+            ++n_tokens;
+        }
+    }
+
+    const auto t_end = std::chrono::high_resolution_clock::now();
+    t_ms = std::chrono::duration_cast<std::chrono::milliseconds>(t_end - t_start).count();
+
+    return result;
+}

data/ext/sources/examples/wchess/libwchess/WChess.h

@@ -0,0 +1,63 @@
+#pragma once
+#include "whisper.h"
+#include <string>
+#include <vector>
+#include <memory>
+
+class Chessboard;
+
+class WChess {
+public:
+    using CheckRunningCb = bool (*)();
+    using GetAudioCb = bool (*)(std::vector<float> &);
+    using SetMovesCb = void (*)(const std::string &, float);
+    using SetGrammarCb = void (*)(const std::string &);
+    using ClearAudioCb = void (*)();
+
+    struct callbacks {
+        GetAudioCb get_audio = nullptr;
+        SetMovesCb set_move = nullptr;
+        SetGrammarCb set_grammar = nullptr;
+    };
+
+    struct settings {
+        int32_t vad_ms     = 2000;
+        int32_t prompt_ms  = 5000;
+        int32_t command_ms = 4000;
+        float vad_thold    = 0.2f;
+        float freq_thold   = 100.0f;
+        bool print_energy  = false;
+    };
+
+    WChess(
+        whisper_context * ctx,
+        const whisper_full_params & wparams,
+        callbacks cb,
+        settings s
+    );
+    ~WChess();
+
+    void run();
+
+    std::string stringify_board() const;
+
+    std::string get_grammar() const;
+
+private:
+    bool get_audio(std::vector<float>& pcmf32) const;
+    void set_move(const std::string& moves, float prob) const;
+    void set_grammar(const std::string& grammar) const;
+
+    std::string transcribe(
+                    const std::vector<float> & pcmf32,
+                    float & logprob_min,
+                    float & logprob_sum,
+                    int & n_tokens,
+                    int64_t & t_ms);
+
+    whisper_context * m_ctx;
+    whisper_full_params m_wparams;
+    const callbacks m_cb;
+    const settings m_settings;
+    std::unique_ptr<Chessboard> m_board;
+};

data/ext/sources/examples/wchess/libwchess/test-chessboard.cpp

@@ -0,0 +1,117 @@
+#include "Chessboard.h"
+
+#define ASSERT(x) \
+    do { \
+        if (!(x)) { \
+            fprintf(stderr, "ASSERT: %s:%d: %s\n", __FILE__, __LINE__, #x); \
+            fflush(stderr); \
+            exit(1); \
+        } \
+    } while (0)
+
+
+int main() {
+    {
+        Chessboard chess;
+
+        ASSERT(chess.process("pawn to d4") == "d2-d4");
+        ASSERT(chess.process("e5") == "e7-e5");
+        ASSERT(chess.process("c1 h6") == "c1-h6");
+        ASSERT(chess.process("queen h4") == "d8-h4");
+        ASSERT(chess.process("bishop to g5") == "h6-g5");
+        ASSERT(chess.process("bishop to b4") == "f8-b4");
+        ASSERT(chess.process("c4") == "");
+        ASSERT(chess.process("knight c3") == "b1-c3");
+        ASSERT(chess.process("knight c6") == "b8-c6");
+        ASSERT(chess.process("f3") == "");
+    }
+
+    {
+        Chessboard chess;
+
+        ASSERT(chess.process("d4") == "d2-d4");
+        ASSERT(chess.process("e5") == "e7-e5");
+        ASSERT(chess.process("e4") == "e2-e4");
+        ASSERT(chess.process("queen h4") == "d8-h4");
+        ASSERT(chess.process("queen h5") == "d1-h5");
+        ASSERT(chess.process("f5") == "");
+        ASSERT(chess.process("g6") == "g7-g6");
+        ASSERT(chess.process("knight e2") == "g1-e2");
+        ASSERT(chess.process("f5") == "f7-f5");
+        ASSERT(chess.process("knight g3") == "e2-g3");
+        ASSERT(chess.process("g5") == "");
+        ASSERT(chess.process("king e7") == "e8-e7");
+        ASSERT(chess.process("f4") == "f2-f4");
+        ASSERT(chess.process("g5") == "g6-g5");
+    }
+
+    {
+        Chessboard chess;
+
+        ASSERT(chess.process("e4") == "e2-e4");
+        ASSERT(chess.process("c5") == "c7-c5");
+        ASSERT(chess.process("e5") == "e4-e5");
+        ASSERT(chess.process("c4") == "c5-c4");
+        ASSERT(chess.process("e6") == "e5-e6");
+        ASSERT(chess.process("c3") == "c4-c3");
+        ASSERT(chess.process("e7") == "");
+        ASSERT(chess.process("f7") == "e6-f7");
+        ASSERT(chess.process("d2") == "");
+        ASSERT(chess.process("king to f7") == "e8-f7");
+        ASSERT(chess.process("f4") == "f2-f4");
+        ASSERT(chess.process("d2") == "c3-d2");
+        ASSERT(chess.process("f5") == "");
+        ASSERT(chess.process("king to e2") == "e1-e2");
+        ASSERT(chess.process("king to g6") == "f7-g6");
+        ASSERT(chess.process("f5") == "f4-f5");
+        ASSERT(chess.process("e6") == "");
+        ASSERT(chess.process("king to h5") == "g6-h5");
+        ASSERT(chess.process("g4") == "g2-g4");
+        ASSERT(chess.process("king to g5") == "h5-g5");
+        ASSERT(chess.process("h4") == "h2-h4");
+        ASSERT(chess.process("king to h5") == "");
+        ASSERT(chess.process("king to g6") == "");
+        ASSERT(chess.process("king to h6") == "g5-h6");
+        ASSERT(chess.process("bishop to d2") == "c1-d2");
+        ASSERT(chess.process("king to g5") == "");
+        ASSERT(chess.process("g5") == "g7-g5");
+    }
+
+    {
+        Chessboard chess;
+        ASSERT(chess.process("f4") == "f2-f4");
+        ASSERT(chess.process("e5") == "e7-e5");
+        ASSERT(chess.process("g4") == "g2-g4");
+        ASSERT(chess.process("queen to h4") == "d8-h4#");
+        ASSERT(chess.process("knight f3") == "");
+        ASSERT(chess.grammar().empty());
+    }
+
+    {
+        Chessboard chess;
+        ASSERT(chess.process("f4") == "f2-f4");
+        ASSERT(chess.process("e5") == "e7-e5");
+        ASSERT(chess.process("g4") == "g2-g4");
+        ASSERT(chess.process("d5") == "d7-d5");
+        ASSERT(chess.process("g1 f3") == "g1-f3");
+        ASSERT(chess.process("queen to h4") == "d8-h4");
+        ASSERT(!chess.grammar().empty());
+    }
+
+    {
+        Chessboard chess;
+        ASSERT(chess.process("knight c3") == "b1-c3");
+        ASSERT(chess.process("knight c6") == "b8-c6");
+        ASSERT(chess.process("knight b5") == "c3-b5");
+        ASSERT(chess.process("knight f6") == "g8-f6");
+        ASSERT(chess.process("knight d6") == "b5-d6");
+        ASSERT(chess.process("knight d4") == "");
+        ASSERT(chess.process("d6") == "c7-d6");
+        ASSERT(chess.process("e4") == "e2-e4");
+        ASSERT(chess.process("knight d4") == "c6-d4");
+        ASSERT(chess.process("d3") == "d2-d3");
+        ASSERT(chess.process("knight e4") == "f6-e4");
+        ASSERT(chess.process("king to e2") == "");
+        ASSERT(chess.process("king to d2") == "");
+    }
+}

data/ext/sources/examples/wchess/wchess.cmd/CMakeLists.txt

@@ -0,0 +1,8 @@
+if (WHISPER_SDL2)
+    set(TARGET wchess)
+    add_executable(${TARGET} wchess.cmd.cpp)
+
+    include(DefaultTargetOptions)
+
+    target_link_libraries(${TARGET} PRIVATE wchess-core common-sdl ${CMAKE_THREAD_LIBS_INIT})
+endif ()

data/ext/sources/examples/wchess/wchess.cmd/wchess.cmd.cpp

@@ -0,0 +1,249 @@
+// Command line voice assisted chess
+//
+// Speak chess move commands to the microphone.
+// The moves will translated to chessboard positions.
+//
+//
+
+#include "WChess.h"
+#include "common-sdl.h"
+#include <iostream>
+
+#include <memory>
+#include <thread>
+
+// command-line parameters
+struct whisper_params {
+    int32_t n_threads  = std::min(4, (int32_t) std::thread::hardware_concurrency());
+    int32_t prompt_ms  = 5000;
+    int32_t command_ms = 8000;
+    int32_t capture_id = -1;
+    int32_t max_tokens = 32;
+    int32_t audio_ctx  = 0;
+
+    float vad_thold  = 0.6f;
+    float freq_thold = 100.0f;
+
+    float grammar_penalty = 100.0f;
+
+    bool translate     = false;
+    bool print_special = false;
+    bool print_energy  = false;
+    bool no_timestamps = true;
+    bool use_gpu       = true;
+    bool flash_attn    = false;
+
+    std::string language  = "en";
+    std::string model     = "models/ggml-base.en.bin";
+    std::string fname_out;
+    std::string commands;
+    std::string prompt;
+    std::string context;
+    std::string grammar;
+};
+
+void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & params) {
+    fprintf(stderr, "\n");
+    fprintf(stderr, "usage: %s [options]\n", argv[0]);
+    fprintf(stderr, "\n");
+    fprintf(stderr, "options:\n");
+    fprintf(stderr, "  -h,         --help           [default] show this help message and exit\n");
+    fprintf(stderr, "  -t N,       --threads N      [%-7d] number of threads to use during computation\n", params.n_threads);
+    fprintf(stderr, "  -pms N,     --prompt-ms N    [%-7d] prompt duration in milliseconds\n",             params.prompt_ms);
+    fprintf(stderr, "  -cms N,     --command-ms N   [%-7d] command duration in milliseconds\n",            params.command_ms);
+    fprintf(stderr, "  -c ID,      --capture ID     [%-7d] capture device ID\n",                           params.capture_id);
+    fprintf(stderr, "  -mt N,      --max-tokens N   [%-7d] maximum number of tokens per audio chunk\n",    params.max_tokens);
+    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, "  -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");
+    fprintf(stderr, "  -ng,        --no-gpu         [%-7s] disable GPU\n",                                 params.use_gpu ? "false" : "true");
+    fprintf(stderr, "  -fa,        --flash-attn     [%-7s] flash attention during decoding\n",             params.flash_attn ? "true" : "false");
+    fprintf(stderr, "  -l LANG,    --language LANG  [%-7s] spoken language\n",                             params.language.c_str());
+    fprintf(stderr, "  -m FNAME,   --model FNAME    [%-7s] model path\n",                                  params.model.c_str());
+    fprintf(stderr, "  -f FNAME,   --file FNAME     [%-7s] text output file name\n",                       params.fname_out.c_str());
+    fprintf(stderr, "  -cmd FNAME, --commands FNAME [%-7s] text file with allowed commands\n",             params.commands.c_str());
+    fprintf(stderr, "  -p,         --prompt         [%-7s] the required activation prompt\n",              params.prompt.c_str());
+    fprintf(stderr, "  -ctx,       --context        [%-7s] sample text to help the transcription\n",       params.context.c_str());
+    fprintf(stderr, "  --grammar-penalty N          [%-7.1f] scales down logits of nongrammar tokens\n",   params.grammar_penalty);
+    fprintf(stderr, "\n");
+}
+
+bool whisper_params_parse(int argc, char ** argv, whisper_params & params) {
+    for (int i = 1; i < argc; i++) {
+        std::string arg = argv[i];
+
+        if (arg == "-h" || arg == "--help") {
+            whisper_print_usage(argc, argv, params);
+            exit(0);
+        }
+        else if (arg == "-t"   || arg == "--threads")       { params.n_threads     = std::stoi(argv[++i]); }
+        else if (arg == "-pms" || arg == "--prompt-ms")     { params.prompt_ms     = std::stoi(argv[++i]); }
+        else if (arg == "-cms" || arg == "--command-ms")    { params.command_ms    = std::stoi(argv[++i]); }
+        else if (arg == "-c"   || arg == "--capture")       { 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")     { 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 == "-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; }
+        else if (arg == "-ng"  || arg == "--no-gpu")        { params.use_gpu       = false; }
+        else if (arg == "-fa"  || arg == "--flash-attn")    { params.flash_attn    = true; }
+        else if (arg == "-l"   || arg == "--language")      { params.language      = argv[++i]; }
+        else if (arg == "-m"   || arg == "--model")         { params.model         = argv[++i]; }
+        else if (arg == "-f"   || arg == "--file")          { params.fname_out     = argv[++i]; }
+        else if (arg == "-cmd" || arg == "--commands")      { params.commands      = argv[++i]; }
+        else if (arg == "-p"   || arg == "--prompt")        { params.prompt        = argv[++i]; }
+        else if (arg == "-ctx" || arg == "--context")       { params.context       = argv[++i]; }
+        else if (                 arg == "--grammar-penalty") { params.grammar_penalty = std::stof(argv[++i]); }
+        else {
+            fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
+            whisper_print_usage(argc, argv, params);
+            exit(0);
+        }
+    }
+
+    return true;
+}
+
+std::unique_ptr<WChess> g_wchess;
+int g_moveCount = 0;
+void set_move(const std::string & move, float) {
+    if (!move.empty()) {
+        g_moveCount++;
+        fprintf(stdout, "Move: %s\n\n", move.c_str());
+    }
+    else fprintf(stdout, "Move rejected\n\n");
+    fprintf(stdout, "%s\n", g_wchess->stringify_board().c_str());
+    fprintf(stdout, "%s\n", g_moveCount ? "White's turn" : "Black's turn");
+}
+
+audio_async g_audio(30*1000);
+bool g_listening = false;
+std::vector<float> g_pcmf32;
+
+bool read_input() {
+    std::string input;
+    while (true) {
+        fprintf(stdout, "[(l)isten/(p)ause/(q)uit]: ");
+        std::cin >> input;
+        fprintf(stdout, "\n");
+        if (input[0] == 'q') {
+            fprintf(stdout, "Quitting\n");
+            return false;
+        }
+        if (input[0] == 'l') {
+            if (!g_listening) {
+                fprintf(stdout, "Listening\n");
+                g_listening = true;
+                g_pcmf32.clear();
+                g_audio.resume();
+                g_audio.clear();
+            }
+            else fprintf(stdout, "Still listening\n");
+            return true;
+        }
+        else {
+            if (g_listening) {
+                g_listening = false;
+                g_audio.get(0, g_pcmf32);
+                g_audio.pause();
+                fprintf(stdout, "Processing\n");
+            }
+            else fprintf(stdout, "Not listening\n");
+            return true;
+        }
+    }
+    return true;
+}
+
+bool get_audio(std::vector<float> & pcmf32_cur) {
+    if (!read_input()) return false;
+    if (!g_pcmf32.empty()) pcmf32_cur = std::move(g_pcmf32);
+    else pcmf32_cur.clear();
+    return true;
+}
+
+int main(int argc, char ** argv) {
+    whisper_params params;
+
+    if (whisper_params_parse(argc, argv, params) == false) {
+        return 1;
+    }
+
+    if (whisper_lang_id(params.language.c_str()) == -1) {
+        fprintf(stderr, "error: unknown language '%s'\n", params.language.c_str());
+        whisper_print_usage(argc, argv, params);
+        exit(0);
+    }
+
+    // whisper init
+
+    struct whisper_context_params cparams = whisper_context_default_params();
+
+    cparams.use_gpu    = params.use_gpu;
+    cparams.flash_attn = params.flash_attn;
+
+    struct whisper_context * ctx = whisper_init_from_file_with_params(params.model.c_str(), cparams);
+    if (!ctx) {
+        fprintf(stderr, "%s: whisper_init_from_file_with_params() failed!\n", __func__);
+        return 1;
+    }
+
+    // init audio
+
+    if (!g_audio.init(params.capture_id, WHISPER_SAMPLE_RATE)) {
+        fprintf(stderr, "%s: audio.init() failed!\n", __func__);
+        return 1;
+    }
+
+    struct whisper_full_params wparams = whisper_full_default_params(whisper_sampling_strategy::WHISPER_SAMPLING_GREEDY);
+    wparams.offset_ms        = 0;
+    wparams.translate        = false;
+    wparams.no_context       = true;
+    wparams.single_segment   = true;
+    wparams.print_realtime   = false;
+    wparams.print_progress   = false;
+    wparams.print_timestamps = true;
+    wparams.print_special    = false;
+    wparams.no_timestamps    = true;
+
+    wparams.max_tokens       = 32;
+    wparams.audio_ctx        = 768; // partial encoder context for better performance
+
+    wparams.temperature     = 0.0f;
+    wparams.temperature_inc = 2.0f;
+    wparams.greedy.best_of  = 1;
+
+    wparams.beam_search.beam_size = 1;
+
+    wparams.language         = "en";
+
+    wparams.grammar_penalty = 100.0;
+
+    wparams.initial_prompt = params.context.data();
+
+    WChess::callbacks cb;
+    cb.get_audio = get_audio;
+    cb.set_move = set_move;
+
+    WChess::settings s;
+    s.vad_ms = 2000;
+    s.prompt_ms = params.prompt_ms;
+    s.command_ms = params.command_ms;
+    s.vad_thold = params.vad_thold;
+    s.freq_thold = params.freq_thold;
+    s.print_energy = params.print_energy;
+
+    g_wchess.reset(new WChess(ctx, wparams, cb, s));
+    set_move("start", 0);
+    g_wchess->run();
+
+    whisper_print_timings(ctx);
+    whisper_free(ctx);
+
+    return 0;
+}