0.0.29

support elevenlabs options

README.md

@@ -1,2 +1,3 @@
 # speech-utils
-TTS-related speech utilities for jambonz
+TTS-related speech utilities for jambonz.
+

index.js

@@ -3,10 +3,12 @@ const {noopLogger} = require('./lib/utils');
 module.exports = (opts, logger) => {
   logger = logger || noopLogger;
   let client = opts.redis_client;
-  if (!client) {
-    const {client: redisClient} = require('@jambonz/realtimedb-helpers')(opts, logger);
-    client = redisClient;
-  }
+  const {
+    client: redisClient,
+    createHash,
+    retrieveHash
+  } = require('@jambonz/realtimedb-helpers')(opts, logger);
+  client = opts.redis_client || redisClient;
 
   return {
     client,
@@ -15,6 +17,7 @@ module.exports = (opts, logger) => {
     synthAudio: require('./lib/synth-audio').bind(null, client, logger),
     getNuanceAccessToken: require('./lib/get-nuance-access-token').bind(null, client, logger),
     getIbmAccessToken: require('./lib/get-ibm-access-token').bind(null, client, logger),
+    getAwsAuthToken: require('./lib/get-aws-sts-token').bind(null, logger, createHash, retrieveHash),
     getTtsVoices: require('./lib/get-tts-voices').bind(null, client, logger),
   };
 };

lib/get-aws-sts-token.js

@@ -0,0 +1,45 @@
+const { STSClient, GetSessionTokenCommand } = require('@aws-sdk/client-sts');
+const {makeAwsKey, noopLogger} = require('./utils');
+const debug = require('debug')('jambonz:speech-utils');
+const EXPIRY = 3600;
+
+async function getAwsAuthToken(
+  logger,
+  createHash, retrieveHash,
+  awsAccessKeyId, awsSecretAccessKey, awsRegion) {
+  logger = logger || noopLogger;
+  try {
+    const key = makeAwsKey(awsAccessKeyId);
+    const obj = await retrieveHash(key);
+    if (obj) return {...obj, servedFromCache: true};
+
+    /* access token not found in cache, so generate it using STS */
+    const stsClient = new STSClient({
+      region: awsRegion,
+      credentials: {
+        accessKeyId: awsAccessKeyId,
+        secretAccessKey: awsSecretAccessKey,
+      }
+    });
+    const command = new GetSessionTokenCommand({DurationSeconds: EXPIRY});
+    const data = await stsClient.send(command);
+
+    const credentials = {
+      accessKeyId: data.Credentials.AccessKeyId,
+      secretAccessKey: data.Credentials.SecretAccessKey,
+      sessionToken: data.Credentials.SessionToken
+    };
+
+    /* expire 10 minutes before the hour, so we don't lose the use of it during a call */
+    createHash(key, credentials, EXPIRY - 600)
+      .catch((err) => logger.error(err, `Error saving hash for key ${key}`));
+
+    return {...credentials, servedFromCache: false};
+  } catch (err) {
+    debug(err, 'getAwsAuthToken: Error retrieving AWS auth token');
+    logger.error(err, 'getAwsAuthToken: Error retrieving AWS auth token');
+    throw err;
+  }
+}
+
+module.exports = getAwsAuthToken;

lib/synth-audio.js

@@ -76,7 +76,7 @@ const trimTrailingSilence = (buffer) => {
  * the synthesized audio, and a variable indicating whether it was served from cache
  */
 async function synthAudio(client, logger, stats, { account_sid,
-  vendor, language, voice, gender, text, engine, salt, model, credentials, deploymentId, disableTtsCache
+  vendor, language, voice, gender, text, engine, salt, model, credentials, deploymentId, disableTtsCache, options
 }) {
   let audioBuffer;
   let servedFromCache = false;
@@ -194,7 +194,7 @@ async function synthAudio(client, logger, stats, { account_sid,
         audioBuffer = await synthWellSaid(logger, {credentials, stats, language, voice, text, filePath});
         break;
       case 'elevenlabs':
-        audioBuffer = await synthElevenlabs(logger, {credentials, stats, language, voice, text, filePath});
+        audioBuffer = await synthElevenlabs(logger, {credentials, options, stats, language, voice, text, filePath});
         break;
       case 'whisper':
         audioBuffer = await synthWhisper(logger, {credentials, stats, voice, text});
@@ -585,21 +585,25 @@ const synthCustomVendor = async(logger, {credentials, stats, language, voice, te
   }
 };
 
-const synthElevenlabs = async(logger, {credentials, stats, language, voice, text}) => {
-  const {api_key, model_id} = credentials;
+const synthElevenlabs = async(logger, {credentials, options, stats, language, voice, text}) => {
+  const {api_key, model_id, options: credOpts} = credentials;
+  const opts = !!options && Object.keys(options).length !== 0 ? options : JSON.parse(credOpts || '{}');
+  const optimize_streaming_latency = opts.optimize_streaming_latency ?
+    `?optimize_streaming_latency=${opts.optimize_streaming_latency}` : '';
   try {
     const post = bent('https://api.elevenlabs.io', 'POST', 'buffer', {
       'xi-api-key': api_key,
       'Accept': 'audio/mpeg',
       'Content-Type': 'application/json'
     });
-    const mp3 = await post(`/v1/text-to-speech/${voice}`, {
+    const mp3 = await post(`/v1/text-to-speech/${voice}${optimize_streaming_latency}`, {
       text,
       model_id,
       voice_settings: {
         stability: 0.5,
         similarity_boost: 0.5
-      }
+      },
+      ...opts
     });
     return mp3;
   } catch (err) {

lib/utils.js

@@ -43,6 +43,12 @@ function makeIbmKey(apiKey) {
   return `ibm:${hash.digest('hex')}`;
 }
 
+function makeAwsKey(awsAccessKeyId) {
+  const hash = crypto.createHash('sha1');
+  hash.update(awsAccessKeyId);
+  return `aws:${hash.digest('hex')}`;
+}
+
 function makeNuanceKey(clientId, secret, scope) {
   const hash = crypto.createHash('sha1');
   hash.update(`${clientId}:${secret}:${scope}`);
@@ -110,6 +116,7 @@ module.exports = {
   makeSynthKey,
   makeNuanceKey,
   makeIbmKey,
+  makeAwsKey,
   getNuanceAccessToken,
   createNuanceClient,
   createKryptonClient,

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.0.26",
+  "version": "0.0.29",
   "description": "TTS-related speech utilities for jambonz",
   "main": "index.js",
   "author": "Dave Horton",
@@ -25,6 +25,7 @@
   "homepage": "https://github.com/jambonz/speech-utils#readme",
   "dependencies": {
     "@aws-sdk/client-polly": "^3.359.0",
+    "@aws-sdk/client-sts": "^3.458.0",
     "@google-cloud/text-to-speech": "^4.2.1",
     "@grpc/grpc-js": "^1.8.13",
     "@jambonz/realtimedb-helpers": "^0.8.7",
@@ -33,7 +34,7 @@
     "form-urlencoded": "^6.1.0",
     "google-protobuf": "^3.21.2",
     "ibm-watson": "^8.0.0",
-    "microsoft-cognitiveservices-speech-sdk": "^1.31.0",
+    "microsoft-cognitiveservices-speech-sdk": "1.32.0",
     "openai": "^4.16.2",
     "undici": "^5.21.0"
   },

test/aws.js

@@ -0,0 +1,40 @@
+const test = require('tape').test ;
+const config = require('config');
+const opts = config.get('redis');
+const logger = require('pino')({level: 'error'});
+process.on('unhandledRejection', (reason, p) => {
+  console.log('Unhandled Rejection at: Promise', p, 'reason:', reason);
+});
+
+const sleep = (ms) => new Promise((resolve) => setTimeout(resolve, ms));
+
+test('AWS - create and cache auth token', async(t) => {
+  const fn = require('..');
+  const {client, getAwsAuthToken} = fn(opts, logger);
+
+  if (!process.env.AWS_ACCESS_KEY_ID || !process.env.AWS_SECRET_ACCESS_KEY || !process.env.AWS_REGION) {
+      t.pass('skipping AWS auth token tests since no AWS credentials provided');
+      t.end();
+      client.quit();
+      return;
+  }
+  try {
+    let obj = await getAwsAuthToken(process.env.AWS_ACCESS_KEY_ID, process.env.AWS_SECRET_ACCESS_KEY, process.env.AWS_REGION);
+    //console.log({obj}, 'received auth token from AWS');
+    t.ok(obj.sessionToken && !obj.servedFromCache, 'successfullY generated auth token from AWS');
+
+    await sleep(250);
+    obj = await getAwsAuthToken(process.env.AWS_ACCESS_KEY_ID, process.env.AWS_SECRET_ACCESS_KEY, process.env.AWS_REGION);
+    //console.log({obj}, 'received auth token from AWS - second request');
+    t.ok(obj.sessionToken && obj.servedFromCache, 'successfully received access token from cache');
+ 
+    await client.flushall();
+    t.end();
+  }
+  catch (err) {
+    console.error(err);
+    t.end(err);
+  }
+  client.quit();
+});
+

test/index.js

@@ -1,4 +1,7 @@
 require('./docker_start');
 require('./synth');
 require('./list-voices');
+require('./aws');
+require('./ibm');
+require('./nuance');
 require('./docker_stop');

test/synth.js

@@ -459,7 +459,16 @@ test('Elevenlabs speech synth tests', async(t) => {
       vendor: 'elevenlabs',
       credentials: {
         api_key: process.env.ELEVENLABS_API_KEY,
-        model_id: process.env.ELEVENLABS_MODEL_ID
+        model_id: process.env.ELEVENLABS_MODEL_ID,
+        options: JSON.stringify({
+          optimize_streaming_latency: 1,
+          voice_settings: {
+            similarity_boost: 1,
+            stability: 0.8,
+            style: 1,
+            use_speaker_boost: true
+          }
+        })
       },
       language: 'en-US',
       voice: process.env.ELEVENLABS_VOICE_ID,