Improve handling of TTS cache by adding the file extension to the cache key

lib/add-file-to-cache.js

@@ -1,6 +1,7 @@
 const fs = require('fs/promises');
 const {noopLogger, makeSynthKey} = require('./utils');
-const EXPIRES = (process.env.JAMBONES_TTS_CACHE_DURATION_MINS || 4 * 60) * 60; // cache tts for 4 hours
+const {JAMBONES_TTS_CACHE_DURATION_MINS} = require('./config');
+const EXPIRES = JAMBONES_TTS_CACHE_DURATION_MINS;
 
 async function addFileToCache(client, logger, path,
   {account_sid, vendor, language, voice, deploymentId, engine, text}) {

lib/config.js

@@ -0,0 +1,17 @@
+const JAMBONES_TTS_TRIM_SILENCE = process.env.JAMBONES_TTS_TRIM_SILENCE;
+const JAMBONES_DISABLE_TTS_STREAMING = process.env.JAMBONES_DISABLE_TTS_STREAMING;
+
+const JAMBONES_HTTP_PROXY_IP = process.env.JAMBONES_HTTP_PROXY_IP;
+const JAMBONES_HTTP_PROXY_PORT = process.env.JAMBONES_HTTP_PROXY_PORT;
+const JAMBONES_TTS_CACHE_DURATION_MINS = (parseInt(process.env.JAMBONES_TTS_CACHE_DURATION_MINS) || 4 * 60) * 60; // cache tts for 4 hours
+
+const TMP_FOLDER = '/tmp';
+
+module.exports = {
+  JAMBONES_TTS_TRIM_SILENCE,
+  JAMBONES_DISABLE_TTS_STREAMING,
+  JAMBONES_HTTP_PROXY_IP,
+  JAMBONES_HTTP_PROXY_PORT,
+  JAMBONES_TTS_CACHE_DURATION_MINS,
+  TMP_FOLDER
+};

lib/synth-audio.js

@@ -19,7 +19,8 @@ const {
   createNuanceClient,
   createKryptonClient,
   createRivaClient,
-  noopLogger
+  noopLogger,
+  makeFilePath
 } = require('./utils');
 const getNuanceAccessToken = require('./get-nuance-access-token');
 const {
@@ -36,8 +37,13 @@ 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 || 4 * 60) * 60; // cache tts for 4 hours
-const TMP_FOLDER = '/tmp';
+const {
+  JAMBONES_DISABLE_TTS_STREAMING,
+  JAMBONES_HTTP_PROXY_IP,
+  JAMBONES_HTTP_PROXY_PORT,
+  JAMBONES_TTS_CACHE_DURATION_MINS,
+} = require('./config');
+const EXPIRES = JAMBONES_TTS_CACHE_DURATION_MINS;
 const OpenAI = require('openai');
 const getAwsAuthToken = require('./get-aws-sts-token');
 
@@ -149,19 +155,7 @@ async function synthAudio(client, createHash, retrieveHash, logger, stats, { acc
     text
   });
   let filePath;
-  if (['nuance', 'nvidia'].includes(vendor) ||
-    (
-      (process.env.JAMBONES_TTS_TRIM_SILENCE || !process.env.JAMBONES_DISABLE_TTS_STREAMING) &&
-      ['microsoft', 'azure'].includes(vendor)
-    ) ||
-    (
-      !process.env.JAMBONES_DISABLE_TTS_STREAMING &&
-      ['elevenlabs', 'deepgram', 'rimelabs'].includes(vendor)
-    )
-  ) {
-    filePath = `${TMP_FOLDER}/${key.replace('tts:', `tts-${salt || ''}`)}.r8`;
-  }
-  else filePath = `${TMP_FOLDER}/${key.replace('tts:', `tts-${salt || ''}`)}.mp3`;
+  filePath = makeFilePath(vendor, key, salt);
   debug(`synth key is ${key}`);
   let cached;
   if (!disableTtsCache) {
@@ -444,7 +438,7 @@ const synthMicrosoft = async(logger, {
       content = `<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="https://www.w3.org/2001/mstts" xml:lang="${language}"><voice name="${voice}">${words}</voice></speak>`;
       logger.info({content}, 'synthMicrosoft');
     }
-    if (!process.env.JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+    if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
       let params = '';
       params += `{api_key=${apiKey}`;
       params += `,language=${language}`;
@@ -454,8 +448,8 @@ const synthMicrosoft = async(logger, {
       if (region) params += `,region=${region}`;
       if (custom_tts_endpoint) params += `,endpointId=${custom_tts_endpoint}`;
       if (custom_tts_endpoint_url) params += `,endpoint=${custom_tts_endpoint_url}`;
-      if (process.env.JAMBONES_HTTP_PROXY_IP) params += `,http_proxy_ip=${process.env.JAMBONES_HTTP_PROXY_IP}`;
-      if (process.env.JAMBONES_HTTP_PROXY_PORT) params += `,http_proxy_port=${process.env.JAMBONES_HTTP_PROXY_PORT}`;
+      if (JAMBONES_HTTP_PROXY_IP) params += `,http_proxy_ip=${JAMBONES_HTTP_PROXY_IP}`;
+      if (JAMBONES_HTTP_PROXY_PORT) params += `,http_proxy_port=${JAMBONES_HTTP_PROXY_PORT}`;
       params += '}';
       return {
         filePath: `say:${params}${content.replace(/\n/g, ' ')}`,
@@ -484,10 +478,10 @@ const synthMicrosoft = async(logger, {
       SpeechSynthesisOutputFormat.Raw8Khz16BitMonoPcm :
       SpeechSynthesisOutputFormat.Audio16Khz32KBitRateMonoMp3;
 
-    if (process.env.JAMBONES_HTTP_PROXY_IP && process.env.JAMBONES_HTTP_PROXY_PORT) {
+    if (JAMBONES_HTTP_PROXY_IP && JAMBONES_HTTP_PROXY_PORT) {
       logger.debug(
-        `synthMicrosoft: using proxy ${process.env.JAMBONES_HTTP_PROXY_IP}:${process.env.JAMBONES_HTTP_PROXY_PORT}`);
-      speechConfig.setProxy(process.env.JAMBONES_HTTP_PROXY_IP, process.env.JAMBONES_HTTP_PROXY_PORT);
+        `synthMicrosoft: using proxy ${JAMBONES_HTTP_PROXY_IP}:${JAMBONES_HTTP_PROXY_PORT}`);
+      speechConfig.setProxy(JAMBONES_HTTP_PROXY_IP, JAMBONES_HTTP_PROXY_PORT);
     }
     const synthesizer = new SpeechSynthesizer(speechConfig);
 
@@ -673,7 +667,7 @@ const synthElevenlabs = async(logger, {
   const opts = !!options && Object.keys(options).length !== 0 ? options : JSON.parse(credOpts || '{}');
 
   /* default to using the streaming interface, unless disabled by env var OR we want just a cache file */
-  if (!process.env.JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+  if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
     let params = '';
     params += `{api_key=${api_key}`;
     params += ',vendor=elevenlabs';
@@ -726,7 +720,7 @@ const synthPlayHT = async(logger, {
   const opts = !!options && Object.keys(options).length !== 0 ? options : JSON.parse(credOpts || '{}');
 
   /* default to using the streaming interface, unless disabled by env var OR we want just a cache file */
-  if (!process.env.JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+  if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
     let params = '';
     params += `{api_key=${api_key}`;
     params += `,user_id=${user_id}`;
@@ -781,7 +775,7 @@ const synthRimelabs = async(logger, {
   const opts = !!options && Object.keys(options).length !== 0 ? options : JSON.parse(credOpts || '{}');
 
   /* default to using the streaming interface, unless disabled by env var OR we want just a cache file */
-  if (!process.env.JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+  if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
     let params = '';
     params += `{api_key=${api_key}`;
     params += `,model_id=${model_id}`;
@@ -823,7 +817,7 @@ const synthRimelabs = async(logger, {
 const synthWhisper = async(logger, {credentials, stats, voice, text, renderForCaching, disableTtsStreaming}) => {
   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 (!process.env.JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+  if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
     let params = '';
     params += `{api_key=${api_key}`;
     params += `,model_id=${model_id}`;
@@ -862,7 +856,7 @@ const synthWhisper = async(logger, {credentials, stats, voice, text, renderForCa
 
 const synthDeepgram = async(logger, {credentials, stats, model, text, renderForCaching, disableTtsStreaming}) => {
   const {api_key} = credentials;
-  if (!process.env.JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+  if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
     let params = '';
     params += `{api_key=${api_key}`;
     params += ',vendor=deepgram';

lib/utils.js

@@ -7,6 +7,7 @@ const HTTP_TIMEOUT = 5000;
 const NUANCE_AUTH_ENDPOINT = 'tts.api.nuance.com:443';
 const grpc = require('@grpc/grpc-js');
 const formurlencoded = require('form-urlencoded');
+const { JAMBONES_DISABLE_TTS_STREAMING, JAMBONES_TTS_TRIM_SILENCE, TMP_FOLDER } = require('./config');
 
 const debug = require('debug')('jambonz:realtimedb-helpers');
 /**
@@ -19,7 +20,44 @@ const debug = require('debug')('jambonz:realtimedb-helpers');
 function makeSynthKey({account_sid = '', vendor, language, voice, engine = '', text}) {
   const hash = crypto.createHash('sha1');
   hash.update(`${language}:${vendor}:${voice}:${engine}:${text}`);
-  return `tts${account_sid ? (':' + account_sid) : ''}:${hash.digest('hex')}`;
+  const hexHashKey = hash.digest('hex');
+  const accountKey = account_sid ? `:${account_sid}` : '';
+  const extension = getFileExtension(vendor);
+  const key = `tts${accountKey}:${extension}:${hexHashKey}`;
+  return key;
+}
+
+function makeFilePath(vendor, key, salt = '') {
+  const extension = getFileExtension(vendor);
+  return `${TMP_FOLDER}/${key.replace('tts:', `tts-${salt}`)}.${extension}`;
+}
+
+function getFileExtension(vendor) {
+  const mp3Extension = 'mp3';
+  const r8Extension = 'r8';
+
+  switch (vendor) {
+    case 'azure':
+    case 'microsoft':
+      if (!JAMBONES_DISABLE_TTS_STREAMING || JAMBONES_TTS_TRIM_SILENCE) {
+        return r8Extension;
+      } else {
+        return mp3Extension;
+      }
+    case 'deepgram':
+    case 'elevenlabs':
+    case 'rimlabs':
+      if (!JAMBONES_DISABLE_TTS_STREAMING) {
+        return r8Extension;
+      } else {
+        return mp3Extension;
+      }
+    case 'nuance':
+    case 'nvidia':
+      return r8Extension;
+    default:
+      return mp3Extension;
+  }
 }
 
 const noopLogger = {
@@ -123,5 +161,6 @@ module.exports = {
   createRivaClient,
   makeBasicAuthHeader,
   NUANCE_AUTH_ENDPOINT,
-  noopLogger
+  noopLogger,
+  makeFilePath
 };

package.json

@@ -12,6 +12,7 @@
     "test": "NODE_ENV=test node test/ ",
     "coverage": "nyc --reporter html --report-dir ./coverage npm run test",
     "jslint": "eslint index.js lib",
+    "jslint:fix": "eslint --fix '**/*.js'",
     "build": "./build_stubs.sh"
   },
   "repository": {

stubs/nuance/synthesizer_grpc_pb.js

@@ -62,9 +62,9 @@ function deserialize_nuance_tts_v1_UnarySynthesisResponse(buffer_arg) {
 
 //
 // The Synthesizer service offers these functionalities:
-// - GetVoices: Queries the list of available voices, with filters to reduce the search space.  
-// - Synthesize: Synthesizes audio from input text and parameters, and returns an audio stream. 
-// - UnarySynthesize: Synthesizes audio from input text and parameters, and returns a single audio response. 
+// - GetVoices: Queries the list of available voices, with filters to reduce the search space.
+// - Synthesize: Synthesizes audio from input text and parameters, and returns an audio stream.
+// - UnarySynthesize: Synthesizes audio from input text and parameters, and returns a single audio response.
 var SynthesizerService = exports.SynthesizerService = {
   getVoices: {
     path: '/nuance.tts.v1.Synthesizer/GetVoices',

stubs/riva/proto/riva_audio_grpc_pb.js

@@ -1 +1 @@
-// GENERATED CODE -- NO SERVICES IN PROTO
+// GENERATED CODE -- NO SERVICES IN PROTO

stubs/riva/proto/riva_tts_grpc_pb.js

@@ -57,7 +57,7 @@ function deserialize_nvidia_riva_tts_SynthesizeSpeechResponse(buffer_arg) {
 var RivaSpeechSynthesisService = exports.RivaSpeechSynthesisService = {
   // Used to request text-to-speech from the service. Submit a request containing the
 // desired text and configuration, and receive audio bytes in the requested format.
-synthesize: {
+  synthesize: {
     path: '/nvidia.riva.tts.RivaSpeechSynthesis/Synthesize',
     requestStream: false,
     responseStream: false,
@@ -69,9 +69,9 @@ synthesize: {
     responseDeserialize: deserialize_nvidia_riva_tts_SynthesizeSpeechResponse,
   },
   // Used to request text-to-speech returned via stream as it becomes available.
-// Submit a SynthesizeSpeechRequest with desired text and configuration,
-// and receive stream of bytes in the requested format.
-synthesizeOnline: {
+  // Submit a SynthesizeSpeechRequest with desired text and configuration,
+  // and receive stream of bytes in the requested format.
+  synthesizeOnline: {
     path: '/nvidia.riva.tts.RivaSpeechSynthesis/SynthesizeOnline',
     requestStream: false,
     responseStream: true,
@@ -83,7 +83,7 @@ synthesizeOnline: {
     responseDeserialize: deserialize_nvidia_riva_tts_SynthesizeSpeechResponse,
   },
   // Enables clients to request the configuration of the current Synthesize service, or a specific model within the service.
-getRivaSynthesisConfig: {
+  getRivaSynthesisConfig: {
     path: '/nvidia.riva.tts.RivaSpeechSynthesis/GetRivaSynthesisConfig',
     requestStream: false,
     responseStream: false,