change to PollyClient

fix: custom tts support multiple audio types

lib/synth-audio.js

@@ -32,7 +32,7 @@ const {
 const {SynthesizeSpeechRequest} = require('../stubs/riva/proto/riva_tts_pb');
 const {AudioEncoding} = require('../stubs/riva/proto/riva_audio_pb');
 const debug = require('debug')('jambonz:realtimedb-helpers');
-const EXPIRES = process.env.JAMBONES_TTS_CACHE_DURATION_MINS || 3600 * 24; // cache tts for 24 hours
+const EXPIRES = (process.env.JAMBONES_TTS_CACHE_DURATION_MINS || 24 * 60) * 60; // cache tts for 24 hours
 const TMP_FOLDER = '/tmp';
 
 /**
@@ -155,7 +155,8 @@ async function synthAudio(client, logger, stats, { account_sid,
         audioBuffer = await synthWellSaid(logger, {credentials, stats, language, voice, text, filePath});
         break;
       case vendor.startsWith('custom') ? vendor : 'cant_match_value':
-        audioBuffer = await synthCustomVendor(logger, {credentials, stats, language, voice, text});
+        ({ audioBuffer, filePath } = await synthCustomVendor(logger,
+          {credentials, stats, language, voice, text, filePath}));
         break;
       default:
         assert(`synthAudio: unsupported speech vendor ${vendor}`);
@@ -183,7 +184,7 @@ async function synthAudio(client, logger, stats, { account_sid,
 
 const synthPolly = async(logger, {credentials, stats, language, voice, engine, text}) => {
   try {
-    const polly = new PollyClient(credentials);
+    const polly = new PollyClient({...credentials});
     const opts = {
       Engine: engine,
       OutputFormat: 'mp3',
@@ -440,30 +441,56 @@ const synthNvidia = async(client, logger, {credentials, stats, language,  voice,
 };
 
 
-// CustomVendor accept only mp3
+const synthCustomVendor = async(logger, {credentials, stats, language, voice, text, filePath}) => {
-const synthCustomVendor = async(logger, {credentials, stats, language, voice, text}) => {
   const {vendor, auth_token, custom_tts_url} = credentials;
 
   try {
-    const post = bent('POST', 'buffer', {
+    const post = bent('POST', {
       'Authorization': `Bearer ${auth_token}`,
-      'Accept': 'audio/mpeg',
       'Content-Type': 'application/json'
     });
 
-    const mp3 = await post(custom_tts_url, {
+    const response = await post(custom_tts_url, {
       language,
-      format: 'audio/mpeg',
       voice,
       type: text.startsWith('<speak>') ? 'ssml' : 'text',
       text
     });
 
-    return mp3;
+    const regex = /\.[^\.]*$/g;
+    const mime = response.headers['content-type'];
+    const buffer = await response.arrayBuffer();
+    return {
+      audioBuffer: buffer,
+      filePath: filePath.replace(regex, getFileExtFromMime(mime))
+    };
   } catch (err) {
     logger.info({err}, `Vendor ${vendor} returned error`);
     throw err;
   }
 };
 
+const getFileExtFromMime = (mime) => {
+  switch (mime) {
+    case 'audio/wav':
+    case 'audio/x-wav':
+      return '.wav';
+    case /audio\/l16.*rate=8000/.test(mime) ? mime : 'cant match value':
+      return '.r8';
+    case /audio\/l16.*rate=16000/.test(mime) ? mime : 'cant match value':
+      return '.r16';
+    case /audio\/l16.*rate=24000/.test(mime) ? mime : 'cant match value':
+      return '.r24';
+    case /audio\/l16.*rate=32000/.test(mime) ? mime : 'cant match value':
+      return '.r32';
+    case /audio\/l16.*rate=48000/.test(mime) ? mime : 'cant match value':
+      return '.r48';
+    case 'audio/mpeg':
+    case 'audio/mp3':
+      return '.mp3';
+    default:
+      return '.wav';
+  }
+};
+
 module.exports = synthAudio;

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.0.2",
+  "version": "0.0.5",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "@jambonz/speech-utils",
-      "version": "0.0.2",
+      "version": "0.0.5",
       "license": "MIT",
       "dependencies": {
         "@aws-sdk/client-polly": "^3.276.0",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.0.2",
+  "version": "0.0.5",
   "description": "TTS-related speech utilities for jambonz",
   "main": "index.js",
   "author": "Dave Horton",

test/synth.js

@@ -348,7 +348,6 @@ test('Custom Vendor speech synth tests', async(t) => {
     let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/somethingnew`);
     t.ok(obj.headers.Authorization == 'Bearer some_jwt_token', 'Custom Vendor Authentication Header is correct');
     t.ok(obj.body.language == 'en-US', 'Custom Vendor Language is correct');
-    t.ok(obj.body.format == 'audio/mpeg', 'Custom Vendor format is correct');
     t.ok(obj.body.voice == 'English-US.Female-1', 'Custom Vendor voice is correct');
     t.ok(obj.body.type == 'text', 'Custom Vendor type is correct');
     t.ok(obj.body.text == 'This is a test.  This is only a test', 'Custom Vendor text is correct');