0.2.20

add playback_id to say metadata

lib/synth-audio.js

@@ -53,6 +53,8 @@ const EXPIRES = JAMBONES_TTS_CACHE_DURATION_MINS;
 const OpenAI = require('openai');
 const getAwsAuthToken = require('./get-aws-sts-token');
 
+/* counter for playback id */
+let playbackIdCounter = 0;
 
 const trimTrailingSilence = (buffer) => {
   assert.ok(buffer instanceof Buffer, 'trimTrailingSilence - argument is not a Buffer');
@@ -211,7 +213,7 @@ async function synthAudio(client, createHash, retrieveHash, logger, stats, { acc
       case 'polly':
         vendorLabel = 'aws';
         audioData = await synthPolly(createHash, retrieveHash, logger,
-          {credentials, stats, language, voice, text, engine});
+          {credentials, stats, language, voice, text, engine, renderForCaching, disableTtsStreaming});
         break;
       case 'azure':
       case 'microsoft':
@@ -307,9 +309,42 @@ async function synthAudio(client, createHash, retrieveHash, logger, stats, { acc
 }
 
 const synthPolly = async(createHash, retrieveHash, logger,
-  {credentials, stats, language, voice, engine, text}) => {
+  {credentials, stats, language, voice, engine, text, renderForCaching, disableTtsStreaming}) => {
+  const {region, accessKeyId, secretAccessKey, roleArn} = credentials;
+  if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+
+    let params = '{';
+    params += `language=${language}`;
+    params += ',write_cache_file=1';
+    params += ',vendor=aws';
+    if (accessKeyId && secretAccessKey) {
+      if (accessKeyId) params += `,accessKeyId=${accessKeyId}`;
+      if (secretAccessKey) params += `,secretAccessKey=${secretAccessKey}`;
+    } else if (roleArn) {
+      const cred = await getAwsAuthToken(
+        logger, createHash, retrieveHash,
+        {
+          region,
+          roleArn
+        });
+
+      if (cred) {
+        params += `,accessKeyId=${cred.accessKeyId}`;
+        params += `,secretAccessKey=${cred.secretAccessKey}`;
+        params += `,sessionToken=${cred.sessionToken}`;
+      }
+    }
+    if (region) params += `,region=${region}`;
+    if (engine) params += `,engine=${engine}`;
+    params += '}';
+
+    return {
+      filePath: `say:${params}${text.replace(/\n/g, ' ').replace(/\r/g, ' ')}`,
+      servedFromCache: false,
+      rtt: 0
+    };
+  }
   try {
-    const {region, accessKeyId, secretAccessKey, roleArn} = credentials;
     let polly;
     if (accessKeyId && secretAccessKey) {
       polly = new PollyClient({
@@ -556,8 +591,8 @@ const synthMicrosoft = async(logger, {
     }
     if (!JAMBONES_DISABLE_TTS_STREAMING && !JAMBONES_DISABLE_AZURE_TTS_STREAMING &&
       !renderForCaching && !disableTtsStreaming) {
-      let params = '';
-      params += `{api_key=${apiKey}`;
+      let params = '{';
+      params += `api_key=${apiKey}`;
       params += `,language=${language}`;
       params += ',vendor=microsoft';
       params += `,voice=${voice}`;
@@ -820,8 +855,9 @@ const synthElevenlabs = async(logger, {
 
   /* default to using the streaming interface, unless disabled by env var OR we want just a cache file */
   if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
-    let params = '';
-    params += `{api_key=${api_key}`;
+    let params = '{';
+    params += `api_key=${api_key}`;
+    params += `,playback_id=${playbackIdCounter++}`;
     params += ',vendor=elevenlabs';
     params += `,voice=${voice}`;
     params += `,model_id=${model_id}`;
@@ -910,8 +946,9 @@ const synthPlayHT = async(client, logger, {
 
   /* default to using the streaming interface, unless disabled by env var OR we want just a cache file */
   if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
-    let params = '';
-    params += `{api_key=${api_key}`;
+    let params = '{';
+    params += `api_key=${api_key}`;
+    params += `,playback_id=${playbackIdCounter++}`;
     params += `,user_id=${user_id}`;
     params += ',vendor=playht';
     params += `,voice=${voice}`;
@@ -977,8 +1014,9 @@ const synthInworld = async(logger, {
 
   /* default to using the streaming interface, unless disabled by env var OR we want just a cache file */
   if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
-    let params = '';
-    params += `{api_key=${api_key}`;
+    let params = '{';
+    params += `api_key=${api_key}`;
+    params += `,playback_id=${playbackIdCounter++}`;
     params += `,model_id=${model_id}`;
     params += ',vendor=inworld';
     params += `,voice=${voice}`;
@@ -1041,8 +1079,9 @@ const synthRimelabs = async(logger, {
 
   /* default to using the streaming interface, unless disabled by env var OR we want just a cache file */
   if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
-    let params = '';
-    params += `{api_key=${api_key}`;
+    let params = '{';
+    params += `api_key=${api_key}`;
+    params += `,playback_id=${playbackIdCounter++}`;
     params += `,model_id=${model_id}`;
     params += ',vendor=rimelabs';
     params += `,language=${language}`;
@@ -1097,8 +1136,9 @@ const synthVerbio = async(client, logger, {credentials, stats, voice, text, rend
   }
   const token = await getVerbioAccessToken(client, logger, credentials);
   if (!process.env.JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
-    let params = '';
-    params += `{access_token=${token.access_token}`;
+    let params = '{';
+    params += `access_token=${token.access_token}`;
+    params += `,playback_id=${playbackIdCounter++}`;
     params += ',vendor=verbio';
     params += `,voice=${voice}`;
     params += ',write_cache_file=1';
@@ -1140,8 +1180,9 @@ const synthWhisper = async(logger, {credentials, stats, voice, text, instruction
   const {api_key, model_id, baseURL, timeout, speed} = credentials;
   /* if the env is set to stream then bag out, unless we are specifically rendering to generate a cache file */
   if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
-    let params = '';
-    params += `{api_key=${api_key}`;
+    let params = '{';
+    params += `api_key=${api_key}`;
+    params += `,playback_id=${playbackIdCounter++}`;
     params += `,model_id=${model_id}`;
     params += ',vendor=whisper';
     params += `,voice=${voice}`;
@@ -1186,8 +1227,9 @@ const synthWhisper = async(logger, {credentials, stats, voice, text, instruction
 const synthDeepgram = async(logger, {credentials, stats, model, text, renderForCaching, disableTtsStreaming}) => {
   const {api_key, deepgram_tts_uri} = credentials;
   if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
-    let params = '';
-    params += `{api_key=${api_key}`;
+    let params = '{';
+    params += `api_key=${api_key}`;
+    params += `,playback_id=${playbackIdCounter++}`;
     params += ',vendor=deepgram';
     params += `,voice=${model}`;
     params += ',write_cache_file=1';
@@ -1229,8 +1271,9 @@ const synthCartesia = async(logger, {
   const opts = !!options && Object.keys(options).length !== 0 ? options : JSON.parse(credOpts || '{}');
 
   if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
-    let params = '';
-    params += `{api_key=${api_key}`;
+    let params = '{';
+    params += `api_key=${api_key}`;
+    params += `,playback_id=${playbackIdCounter++}`;
     params += `,model_id=${model_id}`;
     params += ',vendor=cartesia';
     params += `,voice=${voice}`;
@@ -1295,19 +1338,21 @@ const synthCartesia = async(logger, {
 const synthResemble = async(logger, {
   credentials, options, stats, voice, text, renderForCaching, disableTtsStreaming
 }) => {
-  const {api_key, resemble_tts_uri} = credentials;
+  const {api_key, resemble_tts_uri, resemble_tts_use_tls} = credentials;
   const {project_uuid, use_hd} = options || {};
 
   /* default to using the streaming interface, unless disabled by env var OR we want just a cache file */
   if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
-    let params = '';
-    params += `{api_key=${api_key}`;
+    let params = '{';
+    params += `api_key=${api_key}`;
+    params += `,playback_id=${playbackIdCounter++}`;
     params += ',vendor=resemble';
     params += `,voice=${voice}`;
     params += ',write_cache_file=1';
     if (project_uuid) params += `,project_uuid=${project_uuid}`;
     if (use_hd) params += `,use_hd=${use_hd}`;
     if (resemble_tts_uri) params += `,endpoint=${resemble_tts_uri}`;
+    if (resemble_tts_use_tls) params += `,use_tls=${resemble_tts_use_tls}`;
 
     params += '}';
 

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.2.16",
+  "version": "0.2.20",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "@jambonz/speech-utils",
-      "version": "0.2.16",
+      "version": "0.2.20",
       "license": "MIT",
       "dependencies": {
         "23": "^0.0.0",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.2.16",
+  "version": "0.2.20",
   "description": "TTS-related speech utilities for jambonz",
   "main": "index.js",
   "author": "Dave Horton",

test/synth.js

@@ -185,6 +185,7 @@ test('AWS speech synth tests', async(t) => {
       language: 'en-US',
       voice: 'Joey',
       text: 'This is a test.  This is only a test',
+      renderForCaching: true,
     });
     t.ok(!opts.servedFromCache, `successfully synthesized aws audio to ${opts.filePath}`);
 
@@ -198,6 +199,7 @@ test('AWS speech synth tests', async(t) => {
       language: 'en-US',
       voice: 'Joey',
       text: 'This is a test.  This is only a test',
+      renderForCaching: true,
     });
     t.ok(opts.servedFromCache, `successfully retrieved aws audio from cache ${opts.filePath}`);
   } catch (err) {