BackgroundAudioSpeech.h

/*
    BackgroundAudio
    Plays an audio file using IRQ driven decompression.  Main loop() writes
    data to the buffer but isn't blocked while playing
 
    Copyright (c) 2024 Earle F. Philhower, III <earlephilhower@yahoo.com>
 
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.
 
    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.
 
    You should have received a copy of the GNU General Public License
    along with this program.  If not, see <http://www.gnu.org/licenses/>.
*/
 
#pragma once
#include <Arduino.h>
#include "WrappedAudioOutputBase.h"
#include "BackgroundAudioGain.h"
#include "BackgroundAudioBuffers.h"
#include "libespeak-ng/espeak-ng/speak_lib.h"
#include "libespeak-ng/phoneme/phonindex.h"
#include "libespeak-ng/phoneme/phontab.h"
#include "libespeak-ng/phoneme/phondata.h"
#include "libespeak-ng/phoneme/intonations.h"
 
// These will be defined by including a language
extern const unsigned char __espeakng_dict[];
extern size_t __espeakng_dictlen;
 
typedef struct {
    const char *name;
    size_t len; // Size of binary data
    const unsigned char *data;
} BackgroundAudioVoice;
 
 
template<class DataBuffer>
class BackgroundAudioSpeechClass {
public:
    BackgroundAudioSpeechClass() {
        _playing = false;
        _out = nullptr;
        _paused = false;
        _voice = nullptr;
    }
 
    BackgroundAudioSpeechClass(AudioOutputBase &d) {
        _playing = false;
        _paused = false;
        setDevice(&d);
    }
 
    ~BackgroundAudioSpeechClass() {}
 
    bool setDevice(AudioOutputBase *d) {
        if (!_playing) {
            _out = d;
            return true;
        }
        return false;
    }
 
    void setGain(float scale) {
        _gain = (int32_t)(scale * (1 << 16));
    }
 
    void setVoice(BackgroundAudioVoice &v) {
        _voice = v.data;
        _voiceLen = v.len;
    }
 
    void setRate(int rate) {
        espeak_SetParameter(espeakRATE, rate, 0);
    }
 
    void setPitch(int pitch) {
        espeak_SetParameter(espeakPITCH, pitch, 0);
    }
 
    void setWordGap(int gap) {
        espeak_SetParameter(espeakWORDGAP, gap, 0);
    }
 
    bool begin() {
        if (_playing || !_voice || !_voiceLen) {
            return false;
        }
 
        espeak_EnableSingleStep();
        espeak_InstallDict(__espeakng_dict, __espeakng_dictlen);
        espeak_InstallPhonIndex(_phonindex, sizeof(_phonindex));
        espeak_InstallPhonTab(_phontab, sizeof(_phontab));
        espeak_InstallPhonData(_phondata, sizeof(_phondata));
        espeak_InstallIntonations(_intonations, sizeof(_intonations));
        espeak_InstallVoice(_voice, _voiceLen);
 
        int samplerate = espeak_Initialize(AUDIO_OUTPUT_SYNCH_PLAYBACK, 20, nullptr, 0);
        espeak_SetVoiceByFile("INTERNAL");
        espeak_SetSynthCallback(_speechCB);
 
        // We will use natural frame size to minimize mismatch
        _out->setBuffers(5, framelen);
        _out->onTransmit(&_cb, (void *)this); // The pump we will use to generate our audio
        _out->setBitsPerSample(16);
        _out->setStereo(true);
        _out->setFrequency(samplerate);
        _out->begin();
 
        // Stuff with silence to start
        uint16_t zeros[32] __attribute__((aligned(4))) = {};
        while (_out->availableForWrite() > 32) {
            _out->write((uint8_t *)zeros, sizeof(zeros));
        }
 
        _playing = true;
 
        return true;
    }
 
    void end() {
        if (_playing) {
            _out->end();
        }
        _playing = false;
    }
 
    bool playing() {
        return _playing;
    }
 
    size_t write(const void *data, size_t len) {
        return _ib.write((const uint8_t *)data, len);
    }
 
    size_t speak(const char *string) {
        if (availableForWrite() < strlen(string)) {
            return 0; // All or nothing
        }
        return write((const void *)string, strlen(string) + 1);
    }
 
    size_t speak(const String &string) {
        return speak(string.c_str());
    }
 
 
    size_t availableForWrite() {
        return _ib.availableForWrite();
    }
 
    size_t available() {
        return _ib.available();
    }
 
    bool done() {
        return !available() && !_generatingSpeech;
    }
 
    uint32_t frames() {
        return _frames;
    }
 
    uint32_t shifts() {
        return _shifts;
    }
 
    uint32_t underflows() {
        return _underflows;
    }
 
    uint32_t errors() {
        return _errors;
    }
 
    uint32_t dumps() {
        return _dumps;
    }
 
    void pause() {
        _paused = true;
    }
 
    bool paused() {
        return _paused;
    }
 
 
    void unpause() {
        _paused = false;
    }
 
    void flush() {
        noInterrupts();
        _ib.flush();
        _generatingSpeech = false;
        short *mono;
        espeak_SynthesizeOneStep(&mono); // Thrown out
        espeak_AbortSynthesis();
        interrupts();
    }
 
private:
    static void _cb(void *ptr) {
        ((BackgroundAudioSpeechClass*)ptr)->pump();
    }
 
    static int _speechCB(short *data, int count, espeak_EVENT *events) {
        return 0; // Should never really be called by ESpeak internals
    }
 
    void generateOneFrame() {
        _frameLen = 0;
 
        // If we're not currently synthesizng speech, is there another string we can say?
        if (!_generatingSpeech) {
            if (_ib.available()) {
                const uint8_t *b = _ib.buffer();
                for (int i = 0; i < (int)_ib.available(); i++) {
                    if (!b[i]) {
                        espeak_Synth(_ib.buffer(), i, 0, (espeak_POSITION_TYPE)0, 0, espeakCHARS_AUTO, 0, this);
                        _generatingSpeech = true;
                        break;
                    }
                }
            }
        }
 
        if (_generatingSpeech && !_frameLen) {
            // Generate the raw samples
            short *mono;
            _frameLen = std::min(espeak_SynthesizeOneStep(&mono), framelen);
            // Now convert to stereo by duplicating channels, store in frame buffer
            int16_t *ptr = _frame;
            for (int i = 0; i < _frameLen; i++) {
                *ptr++ = *mono;
                *ptr++ = *mono++;
            }
            // Amplify if requested
            ApplyGain(_frame, _frameLen * 2, _gain);
            // Advance synthesis state and check if done
            if (!espeak_SynthesisGenerateNext()) {
                _generatingSpeech = false;
                _ib.shiftUp(strlen((const char *)_ib.buffer()) + 1); // Only shift out the speech once it's done speaking, easier to track
                _shifts++;
            }
        }
    }
 
    void pump() {
        while (_out->availableForWrite() >= (int)framelen) {
            if (!_frameLen && !_paused) {
                generateOneFrame();
            }
            if (_paused || !_frameLen) {
                bzero(_frame, sizeof(_frame));
                _out->write((uint8_t *)_frame, sizeof(_frame));
            } else {
                _frameLen -= _out->write((uint8_t *)_frame, _frameLen * 4) / 4;
            }
        }
    }
 
private:
    AudioOutputBase *_out;
    bool _playing = false;
    bool _paused = false;
    DataBuffer _ib;
    int32_t _gain = 1 << 16;
    bool _generatingSpeech = false;
    static constexpr int framelen = 1324; // From the 22050 normal samplerate and 20 length
    int16_t _frame[framelen * 2]; // Overprovision in case we get a long speech frame
    int _frameLen = 0;
 
    const unsigned char *_dict;
    size_t _dictLen;
    const unsigned char *_voice;
    size_t _voiceLen;
 
    // Quality stats, cumulative
    uint32_t _frames = 0;
    uint32_t _shifts = 0;
    uint32_t _underflows = 0;
    uint32_t _errors = 0;
    uint32_t _dumps = 0;
};
 
 
using BackgroundAudioSpeech = BackgroundAudioSpeechClass<RawDataBuffer<1024>>;