agentgui 1.0.192 → 1.0.193

package/lib/speech.js

@@ -1,7 +1,6 @@
 import { createRequire } from 'module';
 import fs from 'fs';
 import path from 'path';
-import http from 'http';
 import { fileURLToPath } from 'url';
 
 const require = createRequire(import.meta.url);
@@ -12,64 +11,6 @@ const serverSTT = require('webtalk/server-stt');
 const serverTTS = require('webtalk/server-tts');
 
 const EXTRA_VOICE_DIRS = [path.join(ROOT, 'voices')];
-const TTS_PORT = 8787;
-
-const TTS_CACHE_MAX = 10 * 1024 * 1024;
-let cacheBytes = 0;
-const cache = new Map();
-const inflight = new Map();
-
-function resolveVoice(voiceId) {
-  if (!voiceId || voiceId === 'default') return null;
-  return serverTTS.findVoiceFile(voiceId, EXTRA_VOICE_DIRS);
-}
-
-function cachePut(key, buf) {
-  if (cache.has(key)) { cacheBytes -= cache.get(key).length; cache.delete(key); }
-  while (cacheBytes + buf.length > TTS_CACHE_MAX && cache.size > 0) {
-    const oldest = cache.keys().next().value;
-    cacheBytes -= cache.get(oldest).length;
-    cache.delete(oldest);
-  }
-  cache.set(key, buf);
-  cacheBytes += buf.length;
-}
-
-function sendToPocket(text, voicePath) {
-  return new Promise((resolve, reject) => {
-    const boundary = '----PocketTTS' + Date.now();
-    const parts = [];
-    parts.push(`--${boundary}\r\nContent-Disposition: form-data; name="text"\r\n\r\n${text}\r\n`);
-    if (voicePath) {
-      const data = fs.readFileSync(voicePath);
-      const name = path.basename(voicePath);
-      parts.push(`--${boundary}\r\nContent-Disposition: form-data; name="voice_wav"; filename="${name}"\r\nContent-Type: audio/wav\r\n\r\n`);
-      parts.push(data);
-      parts.push('\r\n');
-    }
-    parts.push(`--${boundary}--\r\n`);
-    const body = Buffer.concat(parts.map(p => Buffer.isBuffer(p) ? p : Buffer.from(p)));
-    const req = http.request({
-      hostname: '127.0.0.1', port: TTS_PORT, path: '/tts', method: 'POST',
-      headers: { 'Content-Type': `multipart/form-data; boundary=${boundary}`, 'Content-Length': body.length },
-      timeout: 60000,
-    }, res => {
-      if (res.statusCode !== 200) {
-        let e = '';
-        res.on('data', d => e += d);
-        res.on('end', () => reject(new Error(`pocket-tts HTTP ${res.statusCode}: ${e}`)));
-        return;
-      }
-      const chunks = [];
-      res.on('data', d => chunks.push(d));
-      res.on('end', () => resolve(Buffer.concat(chunks)));
-    });
-    req.on('error', reject);
-    req.on('timeout', () => { req.destroy(); reject(new Error('pocket-tts timeout')); });
-    req.write(body);
-    req.end();
-  });
-}
 
 function transcribe(audioBuffer) {
   return serverSTT.transcribe(audioBuffer);
@@ -79,37 +20,12 @@ function getSTT() {
   return serverSTT.getSTT();
 }
 
-async function synthesize(text, voiceId) {
-  const status = serverTTS.getStatus();
-  if (!status.ready) throw new Error('pocket-tts not healthy');
-  const key = (voiceId || 'default') + ':' + text;
-  const cached = cache.get(key);
-  if (cached) { cache.delete(key); cache.set(key, cached); return cached; }
-  const existing = inflight.get(key);
-  if (existing) return existing;
-  const promise = (async () => {
-    const voicePath = resolveVoice(voiceId);
-    const wav = await sendToPocket(text, voicePath);
-    if (!wav || wav.length <= 44) throw new Error('pocket-tts returned empty audio');
-    cachePut(key, wav);
-    return wav;
-  })();
-  inflight.set(key, promise);
-  try { return await promise; } finally { inflight.delete(key); }
+function synthesize(text, voiceId) {
+  return serverTTS.synthesize(text, voiceId, EXTRA_VOICE_DIRS);
 }
 
-async function* synthesizeStream(text, voiceId) {
-  const status = serverTTS.getStatus();
-  if (!status.ready) throw new Error('pocket-tts not healthy');
-  const sentences = splitSentences(text);
-  for (const sentence of sentences) {
-    const key = (voiceId || 'default') + ':' + sentence;
-    const cached = cache.get(key);
-    if (cached) { cache.delete(key); cache.set(key, cached); yield cached; continue; }
-    const voicePath = resolveVoice(voiceId);
-    const wav = await sendToPocket(sentence, voicePath);
-    if (wav && wav.length > 44) { cachePut(key, wav); yield wav; }
-  }
+function synthesizeStream(text, voiceId) {
+  return serverTTS.synthesizeStream(text, voiceId, EXTRA_VOICE_DIRS);
 }
 
 function getVoices() {
@@ -142,19 +58,15 @@ function preloadTTS() {
 }
 
 function ttsCacheKey(text, voiceId) {
-  return (voiceId || 'default') + ':' + text;
+  return serverTTS.ttsCacheKey(text, voiceId);
 }
 
 function ttsCacheGet(key) {
-  const cached = cache.get(key);
-  if (cached) { cache.delete(key); cache.set(key, cached); }
-  return cached || null;
+  return serverTTS.ttsCacheGet(key);
 }
 
 function splitSentences(text) {
-  const raw = text.match(/[^.!?]+[.!?]+[\s]?|[^.!?]+$/g);
-  if (!raw) return [text];
-  return raw.map(s => s.trim()).filter(s => s.length > 0);
+  return serverTTS.splitSentences(text);
 }
 
 export { transcribe, synthesize, synthesizeStream, getSTT, getStatus, getVoices, preloadTTS, ttsCacheKey, ttsCacheGet, splitSentences };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agentgui",
-  "version": "1.0.192",
+  "version": "1.0.193",
   "description": "Multi-agent ACP client with real-time communication",
   "type": "module",
   "main": "server.js",