// TTS Engine wrapper with queue
import { MsEdgeTTS, OUTPUT_FORMAT } from 'msedge-tts';
import { mkdir, writeFile } from 'fs/promises';
import { existsSync } from 'fs';
import { join, dirname } from 'path';
import { fileURLToPath } from 'url';
import { AudioQueue } from './audioQueue.js';
import { VOICES, EMOTIONS, DEFAULT_VOICE } from './voiceConfig.js';
// Get the directory of this script
const __filename = fileURLToPath(import.meta.url);
const __dirname = dirname(__filename);
export class VoiceEngine {
constructor() {
this.tts = new MsEdgeTTS();
this.currentVoice = DEFAULT_VOICE;
this.tempDir = join(dirname(dirname(__dirname)), 'temp_voice');
this.audioQueue = new AudioQueue();
}
async initialize() {
if (!existsSync(this.tempDir)) {
await mkdir(this.tempDir, { recursive: true });
}
await this.setVoice(this.currentVoice);
}
async setVoice(voiceId) {
if (!VOICES[voiceId]) {
throw new Error(`Unknown voice: ${voiceId}`);
}
try {
this.currentVoice = voiceId;
await this.tts.setMetadata(
voiceId,
OUTPUT_FORMAT.AUDIO_24KHZ_48KBITRATE_MONO_MP3
);
return true;
} catch (error) {
// Voice still changes despite the error
this.currentVoice = voiceId;
console.error('Voice metadata error (voice still changed):', error.message);
return true;
}
}
async speak(text, emotion = 'neutral') {
try {
const emotionSettings = EMOTIONS[emotion] || EMOTIONS.neutral;
const voiceConfig = VOICES[this.currentVoice];
// Combine base voice pitch with emotion pitch
const basePitch = parseInt(voiceConfig.pitch);
const emotionPitch = parseInt(emotionSettings.pitch);
const combinedPitch = `+${basePitch + emotionPitch}Hz`;
// Generate unique filename for this audio
const timestamp = Date.now();
const randomSuffix = Math.random().toString(36).substring(7);
const uniqueId = `${timestamp}_${randomSuffix}`;
// Create a subdirectory for this specific audio
const audioDir = join(this.tempDir, uniqueId);
await mkdir(audioDir, { recursive: true });
// Generate audio - msedge-tts will create audio.mp3 in the directory
const { audioFilePath } = await this.tts.toFile(
audioDir,
text,
{
rate: emotionSettings.rate,
pitch: combinedPitch,
volume: emotionSettings.volume
}
);
// Create a VBScript that plays audio completely hidden
const vbsScript = `
Set Sound = CreateObject("WMPlayer.OCX.7")
Sound.URL = "${audioFilePath}"
Sound.settings.volume = 100
Sound.settings.setMode "loop", False
Sound.Controls.play
While Sound.playState <> 1
WScript.Sleep 100
Wend
`.trim();
// Write VBScript to temp file with unique name
const vbsPath = join(this.tempDir, `play_${uniqueId}.vbs`);
await writeFile(vbsPath, vbsScript);
// Add to queue instead of playing immediately
this.audioQueue.add(audioFilePath, vbsPath, audioDir);
// Log for debugging
console.error(`Queued: ${uniqueId}/audio.mp3 with voice ${this.currentVoice} (${emotion} emotion)`);
return { success: true, emotion, voice: this.currentVoice, audioFilePath };
} catch (error) {
throw error;
}
}
cleanText(text) {
// Remove markdown and convert emojis
text = text.replace(/[*_`]/g, '');
const replacements = {
'♡': 'heart',
'♥': 'heart',
'★': 'star',
'☆': 'star',
'～': ' ',
'〜': ' ',
'・': ', ',
};
for (const [old, replacement] of Object.entries(replacements)) {
text = text.replace(new RegExp(old, 'g'), replacement);
}
text = text.replace(/\s+/g, ' ').trim();
if (text.length > 500) {
text = text.substring(0, 500) + '...';
}
return text;
}
getCurrentVoice() {
return this.currentVoice;
}
getVoiceInfo(voiceId) {
return VOICES[voiceId];
}
listVoices() {
return Object.entries(VOICES).map(([id, info]) => ({
id,
...info,
note: info.language.startsWith('ja-JP') ? 'Japanese accent voice' : 'Standard voice'
}));
}
}
