fix deepgram and playht has white noise from begining of cache audio

lib/add-file-to-cache.js

@@ -3,29 +3,8 @@ const {noopLogger, makeSynthKey} = require('./utils');
 const {JAMBONES_TTS_CACHE_DURATION_MINS} = require('./config');
 const EXPIRES = JAMBONES_TTS_CACHE_DURATION_MINS;
 
-function getExtensionAndSampleRate(path) {
-  const match = path.match(/\.([^.]*)$/);
-  if (!match) {
-    //default should be wav file.
-    return ['wav', 8000];
-  }
-
-  const extension = match[1];
-  const sampleRateMap = {
-    r8: 8000,
-    r16: 16000,
-    r24: 24000,
-    r44: 44100,
-    r48: 48000,
-    r96: 96000,
-  };
-
-  const sampleRate = sampleRateMap[extension] || 8000;
-  return [extension, sampleRate];
-}
-
 async function addFileToCache(client, logger, path,
-  {account_sid, vendor, language, voice, deploymentId, engine, model, text, instructions}) {
+  {account_sid, vendor, language, voice, deploymentId, engine, text}) {
   let key;
   logger = logger || noopLogger;
 
@@ -36,19 +15,10 @@ async function addFileToCache(client, logger, path,
       language: language || '',
       voice: voice || deploymentId,
       engine,
-      model,
       text,
-      instructions
     });
-    const [extension, sampleRate] = getExtensionAndSampleRate(path);
     const audioBuffer = await fs.readFile(path);
-    await client.setex(key, EXPIRES, JSON.stringify(
-      {
-        audioContent: audioBuffer.toString('base64'),
-        extension,
-        sampleRate
-      }
-    ));
+    await client.setex(key, EXPIRES, audioBuffer.toString('base64'));
   } catch (err) {
     logger.error(err, 'addFileToCache: Error');
     return;

lib/config.js

@@ -2,14 +2,13 @@ const JAMBONES_TTS_TRIM_SILENCE = process.env.JAMBONES_TTS_TRIM_SILENCE;
 const JAMBONES_DISABLE_TTS_STREAMING = process.env.JAMBONES_DISABLE_TTS_STREAMING;
 const JAMBONES_DISABLE_AZURE_TTS_STREAMING = process.env.JAMBONES_DISABLE_AZURE_TTS_STREAMING;
 const JAMBONES_EAGERLY_PRE_CACHE_AUDIO = process.env.JAMBONES_EAGERLY_PRE_CACHE_AUDIO;
-const JAMBONES_AZURE_ENABLE_SSML = process.env.JAMBONES_AZURE_ENABLE_SSML;
 
 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 = process.env.JAMBONES_TMP_FOLDER || '/tmp';
+const TMP_FOLDER = '/tmp';
 
 const HTTP_TIMEOUT = 5000;
 
@@ -22,6 +21,5 @@ module.exports = {
   JAMBONES_TTS_CACHE_DURATION_MINS,
   JAMBONES_EAGERLY_PRE_CACHE_AUDIO,
   TMP_FOLDER,
-  HTTP_TIMEOUT,
-  JAMBONES_AZURE_ENABLE_SSML
+  HTTP_TIMEOUT
 };

lib/get-aws-sts-token.js

@@ -7,23 +7,22 @@ const CACHE_EXPIRY = process.env.AWS_STS_SESSION_RESET_EXPIRY || (EXPIRY - 600);
 
 async function getAwsAuthToken(
   logger, createHash, retrieveHash,
-  {speech_credential_sid, accessKeyId, secretAccessKey, region, roleArn}) {
+  {accessKeyId, secretAccessKey, region, roleArn}) {
   logger = logger || noopLogger;
   try {
-    // if incase instance profile is used, speech_credential_sid will be used as key to lookup cache
-    const key = makeAwsKey(roleArn || accessKeyId || speech_credential_sid);
+    const key = makeAwsKey(roleArn || accessKeyId);
     const obj = await retrieveHash(key);
     if (obj) return {...obj, servedFromCache: true};
-    /* access token not found in cache, so generate it using STS */
+
     let data;
-    let expiry = CACHE_EXPIRY;
     if (roleArn) {
       const stsClient = new STSClient({ region });
       const roleToAssume = { RoleArn: roleArn, RoleSessionName: 'Jambonz_Speech', DurationSeconds: EXPIRY};
       const command = new AssumeRoleCommand(roleToAssume);
 
       data = await stsClient.send(command);
-    } else if (accessKeyId) {
+    } else {
+      /* access token not found in cache, so generate it using STS */
       const stsClient = new STSClient({
         region,
         credentials: {
@@ -33,26 +32,6 @@ async function getAwsAuthToken(
       });
       const command = new GetSessionTokenCommand({DurationSeconds: EXPIRY});
       data = await stsClient.send(command);
-    } else {
-      // instance profile is used.
-      const stsClient = new STSClient({ region });
-      const cred = await stsClient.config.credentials();
-      // method in the AWS SDK automatically fetches credentials using the default credential
-      // provider chain. If the credentials come from an instance profile or an environment
-      // variable, their expiration is controlled by AWS and not explicitly by our code.
-      if (cred && cred.expiration) {
-        const currentTime = new Date();
-        const expiryTime = new Date(cred.expiration);
-        const remainingTimeInSeconds = Math.round((expiryTime - currentTime) / 1000);
-        expiry = remainingTimeInSeconds;
-      }
-      data = {
-        Credentials: {
-          AccessKeyId: cred.accessKeyId,
-          SecretAccessKey: cred.secretAccessKey,
-          SessionToken: cred.sessionToken
-        }
-      };
     }
 
     const credentials = {
@@ -61,11 +40,9 @@ async function getAwsAuthToken(
       sessionToken: data.Credentials.SessionToken,
       securityToken: data.Credentials.SessionToken
     };
-    // Only cache if expiry is good
-    if (expiry > 0) {
-      createHash(key, credentials, expiry)
-        .catch((err) => logger.error(err, `Error saving hash for key ${key}`));
-    }
+
+    createHash(key, credentials, CACHE_EXPIRY)
+      .catch((err) => logger.error(err, `Error saving hash for key ${key}`));
 
     return {...credentials, servedFromCache: false};
   } catch (err) {

lib/purge-tts-cache.js

@@ -12,7 +12,7 @@ const debug = require('debug')('jambonz:realtimedb-helpers');
  * @returns {object} result - {error, purgedCount}
  */
 async function purgeTtsCache(client, logger, {all, account_sid, vendor,
-  language, voice, deploymentId, engine, model, text, instructions} = {all: true}) {
+  language, voice, deploymentId, engine, text} = {all: true}) {
   logger = logger || noopLogger;
 
   let purgedCount = 0, error;
@@ -33,9 +33,7 @@ async function purgeTtsCache(client, logger, {all, account_sid, vendor,
         language: language || '',
         voice: voice || deploymentId,
         engine,
-        model,
         text,
-        instructions
       });
       purgedCount = await client.del(key);
       if (purgedCount === 0) error = 'Specified item not found';

lib/utils.js

@@ -6,7 +6,7 @@ const pool = new Pool('https://auth.crt.nuance.com');
 const NUANCE_AUTH_ENDPOINT = 'tts.api.nuance.com:443';
 const grpc = require('@grpc/grpc-js');
 const formurlencoded = require('form-urlencoded');
-const { TMP_FOLDER, HTTP_TIMEOUT } = require('./config');
+const { JAMBONES_DISABLE_TTS_STREAMING, JAMBONES_TTS_TRIM_SILENCE, TMP_FOLDER, HTTP_TIMEOUT } = require('./config');
 
 const debug = require('debug')('jambonz:realtimedb-helpers');
 /**
@@ -17,27 +17,65 @@ const debug = require('debug')('jambonz:realtimedb-helpers');
 //const nuanceClientMap = new Map();
 
 function makeSynthKey({
-  account_sid = '',
-  vendor,
-  language,
-  voice,
-  engine = '',
-  model = '',
-  text,
-  instructions = '',
-}) {
+  account_sid = '', vendor, language, voice, engine = '', text,
+  renderForCaching = false}) {
   const hash = crypto.createHash('sha1');
-  hash.update(`${language}:${vendor}:${voice}:${engine}:${model}:${text}:${instructions}`);
+  hash.update(`${language}:${vendor}:${voice}:${engine}:${text}`);
   const hexHashKey = hash.digest('hex');
   const accountKey = account_sid ? `:${account_sid}` : '';
-  const key = `tts${accountKey}:${hexHashKey}`;
+  const namespace = vendor.startsWith('custom') ? vendor : getFileExtension({vendor, renderForCaching});
+  const key = `tts${accountKey}:${namespace}:${hexHashKey}`;
   return key;
 }
 
-function makeFilePath({key, salt = '', extension}) {
+function makeFilePath({vendor, key, salt = '', renderForCaching = false}) {
+  const extension = getFileExtension({vendor, renderForCaching});
   return `${TMP_FOLDER}/${key.replace('tts:', `tts-${salt}`)}.${extension}`;
 }
 
+function getFileExtension({vendor, renderForCaching = false}) {
+  const mp3Extension = 'mp3';
+  const wavExtension = 'wav';
+  const r8Extension = 'r8';
+
+  switch (vendor) {
+    case 'azure':
+    case 'microsoft':
+      if (!renderForCaching && !JAMBONES_DISABLE_TTS_STREAMING || JAMBONES_TTS_TRIM_SILENCE) {
+        return r8Extension;
+      } else {
+        return mp3Extension;
+      }
+    case 'elevenlabs':
+    case 'rimelabs':
+      if (renderForCaching || JAMBONES_DISABLE_TTS_STREAMING) {
+        return mp3Extension;
+      } else {
+        return r8Extension;
+      }
+    case 'playht':
+    case 'deepgram':
+      if (renderForCaching || JAMBONES_DISABLE_TTS_STREAMING) {
+        return mp3Extension;
+      } else {
+        return wavExtension;
+      }
+    case 'nuance':
+    case 'nvidia':
+    case 'verbio':
+      return r8Extension;
+    default:
+      // If vendor is custom
+      if (vendor.startsWith('custom')) {
+        if (renderForCaching || JAMBONES_DISABLE_TTS_STREAMING) {
+          return mp3Extension;
+        } else {
+          return r8Extension;
+        }
+      }
+      return mp3Extension;
+  }
+}
 
 const noopLogger = {
   info: () => {},

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.2.22",
+  "version": "0.1.20",
   "description": "TTS-related speech utilities for jambonz",
   "main": "index.js",
   "author": "Dave Horton",
@@ -26,21 +26,19 @@
   },
   "homepage": "https://github.com/jambonz/speech-utils#readme",
   "dependencies": {
-    "23": "^0.0.0",
     "@aws-sdk/client-polly": "^3.496.0",
     "@aws-sdk/client-sts": "^3.496.0",
-    "@cartesia/cartesia-js": "^2.1.0",
-    "@google-cloud/text-to-speech": "^5.5.0",
+    "@google-cloud/text-to-speech": "^5.0.2",
     "@grpc/grpc-js": "^1.9.14",
     "@jambonz/realtimedb-helpers": "^0.8.7",
     "bent": "^7.3.12",
     "debug": "^4.3.4",
     "form-urlencoded": "^6.1.4",
     "google-protobuf": "^3.21.2",
-    "ibm-watson": "^11.0.0",
+    "ibm-watson": "^8.0.0",
     "microsoft-cognitiveservices-speech-sdk": "1.38.0",
-    "openai": "^4.98.0",
-    "undici": "^7.5.0"
+    "openai": "^4.25.0",
+    "undici": "^6.4.0"
   },
   "devDependencies": {
     "config": "^3.3.11",

test/synth.js

@@ -5,7 +5,7 @@ const fs = require('fs');
 const {makeSynthKey} = require('../lib/utils');
 const logger = require('pino')();
 const bent = require('bent');
-const getJSON = bent('json');
+const getJSON = bent('json')
 
 process.on('unhandledRejection', (reason, p) => {
   console.log('Unhandled Rejection at: Promise', p, 'reason:', reason);
@@ -91,17 +91,14 @@ test('Google speech Custom voice synth tests', async(t) => {
   const fn = require('..');
   const {synthAudio, client} = fn(opts, logger);
 
-  if (!process.env.GCP_CUSTOM_VOICE_FILE &&
-    !process.env.GCP_CUSTOM_VOICE_JSON_KEY ||
-    !process.env.GCP_CUSTOM_VOICE_MODEL) {
-    t.pass(`skipping google speech synth tests since neither 
-GCP_CUSTOM_VOICE_FILE nor GCP_CUSTOM_VOICE_JSON_KEY provided, GCP_CUSTOM_VOICE_MODEL is not provided`);
+  if (!process.env.GCP_CUSTOM_VOICE_FILE && !process.env.GCP_CUSTOM_VOICE_JSON_KEY || !process.env.GCP_CUSTOM_VOICE_MODEL) {
+    t.pass('skipping google speech synth tests since neither GCP_CUSTOM_VOICE_FILE nor GCP_CUSTOM_VOICE_JSON_KEY provided, GCP_CUSTOM_VOICE_MODEL is not provided');
     return t.end();
   }
   try {
     const str = process.env.GCP_CUSTOM_VOICE_JSON_KEY || fs.readFileSync(process.env.GCP_CUSTOM_VOICE_FILE);
     const creds = JSON.parse(str);
-    const opts = await synthAudio(stats, {
+    let opts = await synthAudio(stats, {
       vendor: 'google',
       credentials: {
         credentials: {
@@ -112,7 +109,7 @@ GCP_CUSTOM_VOICE_FILE nor GCP_CUSTOM_VOICE_JSON_KEY provided, GCP_CUSTOM_VOICE_M
       language: 'en-AU',
       text: 'This is a test.  This is only a test',
       voice: {
-        reportedUsage: 'REALTIME',
+        reportedUsage:"REALTIME",
         model: process.env.GCP_CUSTOM_VOICE_MODEL
       }
     });
@@ -124,48 +121,6 @@ GCP_CUSTOM_VOICE_FILE nor GCP_CUSTOM_VOICE_JSON_KEY provided, GCP_CUSTOM_VOICE_M
   client.quit();
 });
 
-test('Google speech voice cloning synth tests', async(t) => {
-  const fn = require('..');
-  const {synthAudio, client} = fn(opts, logger);
-
-  if (!process.env.GCP_CUSTOM_VOICE_FILE &&
-    !process.env.GCP_CUSTOM_VOICE_JSON_KEY ||
-    !process.env.GCP_VOICE_CLONING_FILE &&
-    !process.env.GCP_VOICE_CLONING_JSON_KEY) {
-    t.pass(`skipping google speech synth tests since neither 
-GCP_CUSTOM_VOICE_FILE nor GCP_CUSTOM_VOICE_JSON_KEY provided,
-GCP_VOICE_CLONING_FILE nor GCP_VOICE_CLONING_JSON_KEY is not provided`);
-    return t.end();
-  }
-  try {
-    const googleKey = process.env.GCP_CUSTOM_VOICE_JSON_KEY ||
-      fs.readFileSync(process.env.GCP_CUSTOM_VOICE_FILE);
-    const voice_cloning_key = process.env.GCP_VOICE_CLONING_JSON_KEY ||
-      fs.readFileSync(process.env.GCP_VOICE_CLONING_FILE).toString();
-    const creds = JSON.parse(googleKey);
-    const opts = await synthAudio(stats, {
-      vendor: 'google',
-      credentials: {
-        credentials: {
-          client_email: creds.client_email,
-          private_key: creds.private_key,
-          project_id: creds.project_id
-        },
-      },
-      language: 'en-US',
-      text: 'This is a test. This is only a test. This is a test. This is only a test. This is a test. This is only a test',
-      voice: {
-        voice_cloning_key
-      }
-    });
-    t.ok(!opts.servedFromCache, `successfully synthesized google voice cloning audio to ${opts.filePath}`);
-  } catch (err) {
-    console.error(err);
-    t.end(err);
-  }
-  client.quit();
-});
-
 test('AWS speech synth tests', async(t) => {
   const fn = require('..');
   const {synthAudio, client} = fn(opts, logger);
@@ -185,7 +140,6 @@ 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}`);
 
@@ -199,7 +153,6 @@ 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) {
@@ -556,7 +509,6 @@ test('Custom Vendor speech synth tests', async(t) => {
       text: 'This is a test.  This is only a test',
     });
     t.ok(!opts.servedFromCache, `successfully synthesized custom vendor audio to ${opts.filePath}`);
-    t.ok(opts.filePath.endsWith('wav'), 'audio is cached as wav file');
     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');
@@ -564,21 +516,6 @@ test('Custom Vendor speech synth tests', async(t) => {
     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');
 
-    // Checking if cache is stored with wav format
-    opts = await synthAudio(stats, {
-      vendor: 'custom:somethingnew',
-      credentials: {
-        use_for_tts: 1,
-        custom_tts_url: "http://127.0.0.1:3100/somethingnew",
-        auth_token: 'some_jwt_token'
-      },
-      language: 'en-US',
-      voice: 'English-US.Female-1',
-      text: 'This is a test.  This is only a test',
-    });
-    t.ok(opts.servedFromCache, `successfully get custom vendor cached audio to ${opts.filePath}`);
-    t.ok(opts.filePath.endsWith('wav'), 'audio is cached as wav file');
-
     opts = await synthAudio(stats, {
       vendor: 'custom:somethingnew2',
       credentials: {
@@ -688,100 +625,7 @@ test('PlayHT3.0 speech synth tests', async(t) => {
   await testPlayHT(t, 'Play3.0');
 });
 
-test('Cartesia speech synth tests', async(t) => {
-  const fn = require('..');
-  const {synthAudio, client} = fn(opts, logger);
-
-  if (!process.env.CARTESIA_API_KEY) {
-    t.pass('skipping Cartesia speech synth tests since CARTESIA_API_KEY is not provided');
-    return t.end();
-  }
-  const text = 'Hi there and welcome to jambones! ' + Date.now();
-  try {
-    const opts = await synthAudio(stats, {
-      vendor: 'cartesia',
-      credentials: {
-        api_key: process.env.CARTESIA_API_KEY,
-        model_id: 'sonic-english',
-        options: JSON.stringify({
-          speed: 1,
-          emotion: 'female_happy',
-        })
-      },
-      language: 'en',
-      voice: '694f9389-aac1-45b6-b726-9d9369183238',
-      text,
-      renderForCaching: true
-    });
-    t.ok(!opts.servedFromCache, `successfully playht eleven audio to ${opts.filePath}`);
-
-  } catch (err) {
-    console.error(JSON.stringify(err));
-    t.end(err);
-  }
-  client.quit();
-});
-
-test('inworld speech synth', async(t) => {
-  const fn = require('..');
-  const {synthAudio, client} = fn(opts, logger);
-
-  if (!process.env.INWORLD_API_KEY) {
-    t.pass('skipping inworld speech synth tests since INWORLD_API_KEY is not provided');
-    return t.end();
-  }
-  const text = 'Hi there and welcome to jambones!';
-  try {
-    const opts = await synthAudio(stats, {
-      vendor: 'inworld',
-      credentials: {
-        api_key: process.env.INWORLD_API_KEY,
-        model_id: 'inworld-tts-1'
-      },
-      language: 'en',
-      voice: 'Ashley',
-      text,
-      renderForCaching: true
-    });
-    t.ok(!opts.servedFromCache, `successfully synthesized inworld audio to ${opts.filePath}`);
-
-  } catch (err) {
-    console.error(JSON.stringify(err));
-    t.end(err);
-  }
-  client.quit();
-});
-
-test('resemble speech synth', async(t) => {
-  const fn = require('..');
-  const {synthAudio, client} = fn(opts, logger);
-
-  if (!process.env.RESEMBLE_API_KEY) {
-    t.pass('skipping resemble speech synth tests since RESEMBLE_API_KEY is not provided');
-    return t.end();
-  }
-  const text = '<speak prompt="Speak in an excited, upbeat tone">Hello from Resemble!</speak>';
-  try {
-    const opts = await synthAudio(stats, {
-      vendor: 'resemble',
-      credentials: {
-        api_key: process.env.RESEMBLE_API_KEY,
-      },
-      language: 'en',
-      voice: '3f5fb9f1',
-      text,
-      renderForCaching: true
-    });
-    t.ok(!opts.servedFromCache, `successfully synthesized resemble audio to ${opts.filePath}`);
-
-  } catch (err) {
-    console.error(JSON.stringify(err));
-    t.end(err);
-  }
-  client.quit();
-});
-
-test('rimelabs speech synth tests mist', async(t) => {
+test('rimelabs speech synth tests', async(t) => {
   const fn = require('..');
   const {synthAudio, client} = fn(opts, logger);
 
@@ -791,7 +635,7 @@ test('rimelabs speech synth tests mist', async(t) => {
   }
   const text = 'Hi there and welcome to jambones!';
   try {
-    const opts = await synthAudio(stats, {
+    let opts = await synthAudio(stats, {
       vendor: 'rimelabs',
       credentials: {
         api_key: process.env.RIMELABS_API_KEY,
@@ -801,7 +645,7 @@ test('rimelabs speech synth tests mist', async(t) => {
           reduceLatency: false
         })
       },
-      language: 'eng',
+      language: 'en-US',
       voice: 'amber',
       text,
       renderForCaching: true
@@ -815,40 +659,6 @@ test('rimelabs speech synth tests mist', async(t) => {
   client.quit();
 });
 
-test('rimelabs speech synth tests mistv2', async(t) => {
-  const fn = require('..');
-  const {synthAudio, client} = fn(opts, logger);
-
-  if (!process.env.RIMELABS_API_KEY) {
-    t.pass('skipping rimelabs speech synth tests since RIMELABS_API_KEY is not provided');
-    return t.end();
-  }
-  const text = 'Hi there and welcome to jambones!';
-  try {
-    const opts = await synthAudio(stats, {
-      vendor: 'rimelabs',
-      credentials: {
-        api_key: process.env.RIMELABS_API_KEY,
-        model_id: 'mistv2',
-        options: JSON.stringify({
-          speedAlpha: 1.0,
-          reduceLatency: false
-        })
-      },
-      language: 'spa',
-      voice: 'pablo',
-      text,
-      renderForCaching: true
-    });
-    t.ok(!opts.servedFromCache, `successfully synthesized rimelabs mistv2 audio to ${opts.filePath}`);
-
-  } catch (err) {
-    console.error(JSON.stringify(err));
-    t.end(err);
-  }
-  client.quit();
-});
-
 test('whisper speech synth tests', async(t) => {
   const fn = require('..');
   const {synthAudio, client} = fn(opts, logger);
@@ -948,12 +758,11 @@ test('TTS Cache tests', async(t) => {
     // save some random tts keys to cache
     const minRecords = 8;
     for (const i in Array(minRecords).fill(0)) {
-      await client.set(makeSynthKey({vendor: i, language: i, voice: i, engine: i, model: i, text: i,
-        instructions: i}), i);
+      await client.set(makeSynthKey({vendor: i, language: i, voice: i, engine: i, text: i}), i);
     }
     const count = await getTtsSize();
     t.ok(count >= minRecords, 'getTtsSize worked.');
-
+    
     const {purgedCount} = await purgeTtsCache();
     t.ok(purgedCount >= minRecords, `successfully purged at least ${minRecords} tts records from cache`);
 
@@ -968,7 +777,7 @@ test('TTS Cache tests', async(t) => {
   try {
     // save some random tts keys to cache
     for (const i in Array(10).fill(0)) {
-      await client.set(makeSynthKey({vendor: i, language: i, voice: i, engine: i, text: i, instructions: i}), i);
+      await client.set(makeSynthKey({vendor: i, language: i, voice: i, engine: i, text: i}), i);
     }
     // save a specific key to tts cache
     const opts = {vendor: 'aws', language: 'en-US', voice: 'MALE', engine: 'Engine', text: 'Hello World!'};
@@ -984,15 +793,24 @@ test('TTS Cache tests', async(t) => {
       language: 'non-existing',
       voice: 'non-existing',
     });
-    t.ok(purgedCountWhenErrored === 0, 'purged no records when specified key was not found');
-    t.ok(error, 'error returned when specified key was not found');
+    t.ok(purgedCountWhenErrored === 0, `purged no records when specified key was not found`);
+    t.ok(error, `error returned when specified key was not found`);
 
     // make sure other tts keys are still there
-    const cached = await client.keys('tts:*');
-    t.ok(cached.length >= 1, 'successfully kept all non-specified tts records in cache');
+    const cached = await client.keys('tts:*')
+    t.ok(cached.length >= 1, `successfully kept all non-specified tts records in cache`);
 
-    process.env.VG_TRIM_TTS_SILENCE = 'true';
+    // retrieve keys from cache and check the key contains the file extension    
+    let key = cached[0];
+    t.ok(key.includes('mp3'), `tts cache extension shoult be part of the key and equal mp3`);
+
+    process.env.VG_TRIM_TTS_SILENCE = 'true';    
     await client.set(makeSynthKey({ vendor: 'azure' }), 'value');
+
+    const r8Keys = await client.keys('tts:r8*');
+    key = r8Keys[0];
+    t.ok(key.includes('r8'), `tts cache extension shoult be part of the key and equal r8`);
+
   } catch (err) {
     console.error(JSON.stringify(err));
     t.end(err);
@@ -1006,10 +824,10 @@ test('TTS Cache tests', async(t) => {
     const account_sid = "12412512_cabc_5aff"
     const account_sid2 = "22412512_cabc_5aff"
     for (const i in Array(minRecords).fill(0)) {
-      await client.set(makeSynthKey({account_sid, vendor: i, language: i, voice: i, engine: i, text: i, instructions: i}), i);
+      await client.set(makeSynthKey({account_sid, vendor: i, language: i, voice: i, engine: i, text: i}), i);
     }
     for (const i in Array(minRecords).fill(0)) {
-      await client.set(makeSynthKey({account_sid: account_sid2, vendor: i, language: i, voice: i, engine: i, text: i, instructions: i}), i);
+      await client.set(makeSynthKey({account_sid: account_sid2, vendor: i, language: i, voice: i, engine: i, text: i}), i);
     }
     const {purgedCount} = await purgeTtsCache({account_sid});
     t.equal(purgedCount, minRecords, `successfully purged at least ${minRecords} tts records from cache for account_sid:${account_sid}`);