wip

lib/synth-audio.js

@@ -205,7 +205,8 @@ async function synthAudio(client, createHash, retrieveHash, logger, stats, { acc
     switch (vendor) {
       case 'google':
         audioData = await synthGoogle(logger, {
-          credentials, stats, language, voice, gender, key, text, model, options, instructions
+          credentials, stats, language, voice, gender, key, text, model, options, instructions,
+          renderForCaching, disableTtsStreaming, disableTtsCache
         });
         break;
       case 'aws':
@@ -413,12 +414,49 @@ const synthPolly = async(createHash, retrieveHash, logger,
 
 
 const synthGoogle = async(logger, {
-  credentials, stats, language, voice, gender, text, model, options, instructions
+  credentials, stats, language, voice, gender, key, text, model, options, instructions,
+  renderForCaching, disableTtsStreaming, disableTtsCache
 }) => {
-  const client = new ttsGoogle.TextToSpeechClient(credentials);
-
   const isGemini = !!model;
   const isVoiceCloning = typeof voice === 'object' && voice.voice_cloning_key;
+  // HD voices have pattern like en-US-Chirp3-HD-Charon
+  const isHDVoice = typeof voice === 'string' && voice.includes('-HD-');
+  // Live API is used for Gemini TTS and HD voices
+  const useLiveApi = isGemini || isHDVoice;
+
+  // Streaming support for Google TTS (Gemini, HD voices, and standard voices)
+  // Voice cloning does not support streaming
+  if (!isVoiceCloning && !JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+    // Strip SSML tags for Gemini TTS (it doesn't support SSML)
+    let inputText = text;
+    if (isGemini && text.startsWith('<speak>')) {
+      inputText = text.replace(/<[^>]*>/g, '').trim();
+      logger.info('synthGoogle: Gemini TTS does not support SSML, stripped tags from input');
+    }
+
+    let params = '{';
+    params += `credentials=${JSON.stringify(credentials)}`;
+    params += `,playback_id=${key}`;
+    params += ',vendor=google';
+    params += `,voice=${voice}`;
+    params += `,language_code=${language || 'en-US'}`;
+    params += `,write_cache_file=${disableTtsCache ? 0 : 1}`;
+    params += `,use_live_api=${useLiveApi ? 1 : 0}`;
+    if (model) params += `,model_name=${model}`;
+    if (gender) params += `,gender=${gender}`;
+    // comma is used to separate parameters in freeswitch tts module
+    const prompt = options?.prompt || instructions;
+    if (prompt) params += `,prompt=${prompt.replace(/\n/g, ' ').replace(/,/g, ';')}`;
+    params += '}';
+
+    return {
+      filePath: `say:${params}${(isGemini ? inputText : text).replace(/\n/g, ' ')}`,
+      servedFromCache: false,
+      rtt: 0
+    };
+  }
+
+  const client = new ttsGoogle.TextToSpeechClient(credentials);
 
   // Build input based on voice type
   let input;

test/synth.js

@@ -256,6 +256,226 @@ test('Google Gemini TTS synth tests', async(t) => {
   client.quit();
 });
 
+test('Google TTS streaming tests (!JAMBONES_DISABLE_TTS_STREAMING)', async(t) => {
+  // Ensure streaming is enabled (default behavior)
+  delete process.env.JAMBONES_DISABLE_TTS_STREAMING;
+
+  // Clear require cache to reload config with new env var
+  delete require.cache[require.resolve('../lib/config')];
+  delete require.cache[require.resolve('../lib/synth-audio')];
+  delete require.cache[require.resolve('..')];
+
+  const fn = require('..');
+  const {synthAudio, client} = fn(opts, logger);
+
+  if (!process.env.GCP_FILE && !process.env.GCP_JSON_KEY) {
+    t.pass('skipping Google TTS streaming tests since neither GCP_FILE nor GCP_JSON_KEY provided');
+    return t.end();
+  }
+
+  try {
+    const str = process.env.GCP_JSON_KEY || fs.readFileSync(process.env.GCP_FILE);
+    const creds = JSON.parse(str);
+    const geminiModel = process.env.GCP_GEMINI_TTS_MODEL || 'gemini-2.5-flash-tts';
+
+    // Test 1: Standard voice streaming (use_live_api=0)
+    let result = await synthAudio(stats, {
+      vendor: 'google',
+      credentials: {
+        credentials: {
+          client_email: creds.client_email,
+          private_key: creds.private_key,
+        },
+      },
+      language: 'en-US',
+      voice: 'en-US-Wavenet-D',
+      gender: 'MALE',
+      text: 'This is a test of standard voice streaming.',
+      disableTtsCache: true
+    });
+    t.ok(result.filePath.startsWith('say:'), 'Standard voice returns streaming say: path');
+    t.ok(result.filePath.includes('vendor=google'), 'Standard voice streaming path contains vendor=google');
+    t.ok(result.filePath.includes('use_live_api=0'), 'Standard voice uses use_live_api=0');
+    t.ok(result.filePath.includes('voice=en-US-Wavenet-D'), 'Standard voice streaming path contains voice');
+
+    // Test 2: HD voice streaming (use_live_api=1)
+    result = await synthAudio(stats, {
+      vendor: 'google',
+      credentials: {
+        credentials: {
+          client_email: creds.client_email,
+          private_key: creds.private_key,
+        },
+      },
+      language: 'en-US',
+      voice: 'en-US-Chirp3-HD-Charon',
+      text: 'This is a test of HD voice streaming.',
+      disableTtsCache: true
+    });
+    t.ok(result.filePath.startsWith('say:'), 'HD voice returns streaming say: path');
+    t.ok(result.filePath.includes('vendor=google'), 'HD voice streaming path contains vendor=google');
+    t.ok(result.filePath.includes('use_live_api=1'), 'HD voice uses use_live_api=1 (Live API)');
+    t.ok(result.filePath.includes('voice=en-US-Chirp3-HD-Charon'), 'HD voice streaming path contains voice');
+
+    // Test 3: Gemini TTS streaming (use_live_api=1)
+    result = await synthAudio(stats, {
+      vendor: 'google',
+      credentials: {
+        credentials: {
+          client_email: creds.client_email,
+          private_key: creds.private_key,
+        },
+      },
+      language: 'en-US',
+      voice: 'Kore',
+      model: geminiModel,
+      text: 'This is a test of Gemini TTS streaming.',
+      instructions: 'Speak naturally.',
+      disableTtsCache: true
+    });
+    t.ok(result.filePath.startsWith('say:'), 'Gemini TTS returns streaming say: path');
+    t.ok(result.filePath.includes('vendor=google'), 'Gemini TTS streaming path contains vendor=google');
+    t.ok(result.filePath.includes('use_live_api=1'), 'Gemini TTS uses use_live_api=1 (Live API)');
+    t.ok(result.filePath.includes(`model_name=${geminiModel}`), 'Gemini TTS streaming path contains model_name');
+    t.ok(result.filePath.includes('prompt=Speak naturally.'), 'Gemini TTS streaming path contains prompt');
+
+    // Test 4: Gemini TTS with SSML stripping in streaming mode
+    result = await synthAudio(stats, {
+      vendor: 'google',
+      credentials: {
+        credentials: {
+          client_email: creds.client_email,
+          private_key: creds.private_key,
+        },
+      },
+      language: 'en-US',
+      voice: 'Leda',
+      model: geminiModel,
+      text: '<speak>This SSML should be stripped.</speak>',
+      instructions: 'Speak naturally.',
+      disableTtsCache: true
+    });
+    t.ok(result.filePath.startsWith('say:'), 'Gemini TTS with SSML returns streaming say: path');
+    t.ok(!result.filePath.includes('<speak>'), 'SSML tags are stripped from streaming path');
+    t.ok(result.filePath.includes('This SSML should be stripped.'), 'Text content is preserved after SSML stripping');
+
+    // Test 5: Gemini TTS with prompt containing special characters
+    result = await synthAudio(stats, {
+      vendor: 'google',
+      credentials: {
+        credentials: {
+          client_email: creds.client_email,
+          private_key: creds.private_key,
+        },
+      },
+      language: 'en-US',
+      voice: 'Kore',
+      model: geminiModel,
+      text: 'Testing special characters in prompt.',
+      options: { prompt: 'Speak in a warm, friendly tone' },
+      disableTtsCache: true
+    });
+    t.ok(result.filePath.startsWith('say:'), 'Gemini TTS with special chars returns streaming say: path');
+    // Commas in prompt should be replaced with semicolons
+    t.ok(result.filePath.includes('prompt=Speak in a warm; friendly tone'), 'Commas in prompt are escaped to semicolons');
+
+  } catch (err) {
+    console.error(err);
+    t.end(err);
+  }
+  client.quit();
+});
+
+test('Google TTS non-streaming tests (JAMBONES_DISABLE_TTS_STREAMING=true)', async(t) => {
+  // Enable streaming disable flag
+  process.env.JAMBONES_DISABLE_TTS_STREAMING = 'true';
+
+  // Clear require cache to reload config with new env var
+  delete require.cache[require.resolve('../lib/config')];
+  delete require.cache[require.resolve('../lib/synth-audio')];
+  delete require.cache[require.resolve('..')];
+
+  const fn = require('..');
+  const {synthAudio, client} = fn(opts, logger);
+
+  if (!process.env.GCP_FILE && !process.env.GCP_JSON_KEY) {
+    t.pass('skipping Google TTS non-streaming tests since neither GCP_FILE nor GCP_JSON_KEY provided');
+    delete process.env.JAMBONES_DISABLE_TTS_STREAMING;
+    return t.end();
+  }
+
+  try {
+    const str = process.env.GCP_JSON_KEY || fs.readFileSync(process.env.GCP_FILE);
+    const creds = JSON.parse(str);
+    const geminiModel = process.env.GCP_GEMINI_TTS_MODEL || 'gemini-2.5-flash-tts';
+
+    // Test 1: Standard voice falls back to non-streaming API
+    let result = await synthAudio(stats, {
+      vendor: 'google',
+      credentials: {
+        credentials: {
+          client_email: creds.client_email,
+          private_key: creds.private_key,
+        },
+      },
+      language: 'en-US',
+      voice: 'en-US-Wavenet-D',
+      gender: 'MALE',
+      text: 'This is a test with streaming disabled.',
+      disableTtsCache: true
+    });
+    t.ok(!result.filePath.startsWith('say:'), 'Standard voice does NOT return streaming say: path when disabled');
+    t.ok(result.filePath.endsWith('.mp3'), 'Standard voice returns mp3 file path');
+
+    // Test 2: HD voice falls back to non-streaming API
+    result = await synthAudio(stats, {
+      vendor: 'google',
+      credentials: {
+        credentials: {
+          client_email: creds.client_email,
+          private_key: creds.private_key,
+        },
+      },
+      language: 'en-US',
+      voice: 'en-US-Chirp3-HD-Charon',
+      text: 'This is a test of HD voice with streaming disabled.',
+      disableTtsCache: true
+    });
+    t.ok(!result.filePath.startsWith('say:'), 'HD voice does NOT return streaming say: path when disabled');
+    t.ok(result.filePath.endsWith('.mp3'), 'HD voice returns mp3 file path');
+
+    // Test 3: Gemini TTS falls back to non-streaming API
+    result = await synthAudio(stats, {
+      vendor: 'google',
+      credentials: {
+        credentials: {
+          client_email: creds.client_email,
+          private_key: creds.private_key,
+        },
+      },
+      language: 'en-US',
+      voice: 'Kore',
+      model: geminiModel,
+      text: 'This is a test of Gemini TTS with streaming disabled.',
+      instructions: 'Speak naturally.',
+      disableTtsCache: true
+    });
+    t.ok(!result.filePath.startsWith('say:'), 'Gemini TTS does NOT return streaming say: path when disabled');
+    t.ok(result.filePath.endsWith('.mp3'), 'Gemini TTS returns mp3 file path');
+
+  } catch (err) {
+    console.error(err);
+    t.end(err);
+  } finally {
+    // Clean up: restore default behavior
+    delete process.env.JAMBONES_DISABLE_TTS_STREAMING;
+    delete require.cache[require.resolve('../lib/config')];
+    delete require.cache[require.resolve('../lib/synth-audio')];
+    delete require.cache[require.resolve('..')];
+  }
+  client.quit();
+});
+
 test('AWS speech synth tests', async(t) => {
   const fn = require('..');
   const {synthAudio, client} = fn(opts, logger);