version update

allow configure STS session expiry

.husky/pre-commit

@@ -1,3 +1,3 @@
-npm audit
+#npm audit
 npm run jslint:fix || true
 npm test

lib/config.js

@@ -1,5 +1,6 @@
 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_HTTP_PROXY_IP = process.env.JAMBONES_HTTP_PROXY_IP;
 const JAMBONES_HTTP_PROXY_PORT = process.env.JAMBONES_HTTP_PROXY_PORT;
@@ -13,6 +14,7 @@ const HTTP_TIMEOUT = 5000;
 module.exports = {
   JAMBONES_TTS_TRIM_SILENCE,
   JAMBONES_DISABLE_TTS_STREAMING,
+  JAMBONES_DISABLE_AZURE_TTS_STREAMING,
   JAMBONES_HTTP_PROXY_IP,
   JAMBONES_HTTP_PROXY_PORT,
   JAMBONES_TTS_CACHE_DURATION_MINS,

lib/get-aws-sts-token.js

@@ -1,20 +1,22 @@
 const { STSClient, GetSessionTokenCommand, AssumeRoleCommand } = require('@aws-sdk/client-sts');
 const {makeAwsKey, noopLogger} = require('./utils');
 const debug = require('debug')('jambonz:speech-utils');
-const EXPIRY = 3600;
+const EXPIRY = process.env.AWS_STS_SESSION_DURATION || 3600;
+// by default reset aws session before expiry time 10 mins
+const CACHE_EXPIRY = process.env.AWS_STS_SESSION_RESET_EXPIRY || (EXPIRY - 600);
 
 async function getAwsAuthToken(
   logger, createHash, retrieveHash,
-  awsAccessKeyId, awsSecretAccessKey, awsRegion, roleArn = null) {
+  {accessKeyId, secretAccessKey, region, roleArn}) {
   logger = logger || noopLogger;
   try {
-    const key = makeAwsKey(roleArn || awsAccessKeyId);
+    const key = makeAwsKey(roleArn || accessKeyId);
     const obj = await retrieveHash(key);
     if (obj) return {...obj, servedFromCache: true};
 
     let data;
     if (roleArn) {
-      const stsClient = new STSClient({ region: awsRegion});
+      const stsClient = new STSClient({ region });
       const roleToAssume = { RoleArn: roleArn, RoleSessionName: 'Jambonz_Speech', DurationSeconds: EXPIRY};
       const command = new AssumeRoleCommand(roleToAssume);
 
@@ -22,10 +24,10 @@ async function getAwsAuthToken(
     } else {
       /* access token not found in cache, so generate it using STS */
       const stsClient = new STSClient({
-        region: awsRegion,
+        region,
         credentials: {
-          accessKeyId: awsAccessKeyId,
-          secretAccessKey: awsSecretAccessKey,
+          accessKeyId,
+          secretAccessKey,
         }
       });
       const command = new GetSessionTokenCommand({DurationSeconds: EXPIRY});
@@ -39,8 +41,7 @@ async function getAwsAuthToken(
       securityToken: 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)
+    createHash(key, credentials, CACHE_EXPIRY)
       .catch((err) => logger.error(err, `Error saving hash for key ${key}`));
 
     return {...credentials, servedFromCache: false};

lib/get-tts-voices.js

@@ -107,7 +107,12 @@ const getAwsVoices = async(_client, createHash, retrieveHash, logger, credential
     } else if (roleArn) {
       client = new PollyClient({
         region,
-        credentials: await getAwsAuthToken(logger, createHash, retrieveHash, null, null, region, roleArn),
+        credentials: await getAwsAuthToken(
+          logger, createHash, retrieveHash,
+          {
+            region,
+            roleArn
+          }),
       });
     } else {
       client = new PollyClient({region});

lib/synth-audio.js

@@ -40,6 +40,7 @@ const {AudioEncoding} = require('../stubs/riva/proto/riva_audio_pb');
 const debug = require('debug')('jambonz:realtimedb-helpers');
 const {
   JAMBONES_DISABLE_TTS_STREAMING,
+  JAMBONES_DISABLE_AZURE_TTS_STREAMING,
   JAMBONES_HTTP_PROXY_IP,
   JAMBONES_HTTP_PROXY_PORT,
   JAMBONES_TTS_CACHE_DURATION_MINS,
@@ -280,7 +281,12 @@ const synthPolly = async(createHash, retrieveHash, logger,
     } else if (roleArn) {
       polly = new PollyClient({
         region,
-        credentials: await getAwsAuthToken(logger, createHash, retrieveHash, null, null, region, roleArn),
+        credentials: await getAwsAuthToken(
+          logger, createHash, retrieveHash,
+          {
+            region,
+            roleArn
+          }),
       });
     } else {
       // AWS RoleArn assigned to Instance profile
@@ -443,7 +449,8 @@ 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}">${words}</voice></speak>`;
       logger.info({content}, 'synthMicrosoft');
     }
-    if (!JAMBONES_DISABLE_TTS_STREAMING && !renderForCaching && !disableTtsStreaming) {
+    if (!JAMBONES_DISABLE_TTS_STREAMING && !JAMBONES_DISABLE_AZURE_TTS_STREAMING &&
+      !renderForCaching && !disableTtsStreaming) {
       let params = '';
       params += `{api_key=${apiKey}`;
       params += `,language=${language}`;
@@ -684,6 +691,8 @@ const synthElevenlabs = async(logger, {
     if (opts.voice_settings?.stability) params += `,stability=${opts.voice_settings.stability}`;
     if (opts.voice_settings?.style) params += `,style=${opts.voice_settings.style}`;
     if (opts.voice_settings?.use_speaker_boost === false) params += ',use_speaker_boost=false';
+    if (opts.previous_text) params += `,previous_text=${opts.previous_text}`;
+    if (opts.next_text) params += `,next_text=${opts.next_text}`;
     params += '}';
 
     return {

lib/utils.js

@@ -53,6 +53,7 @@ function getFileExtension(vendor) {
       }
     case 'nuance':
     case 'nvidia':
+    case 'verbio':
       return r8Extension;
     default:
       return mp3Extension;

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.1.4",
+  "version": "0.1.12",
   "description": "TTS-related speech utilities for jambonz",
   "main": "index.js",
   "author": "Dave Horton",
@@ -36,7 +36,7 @@
     "form-urlencoded": "^6.1.4",
     "google-protobuf": "^3.21.2",
     "ibm-watson": "^8.0.0",
-    "microsoft-cognitiveservices-speech-sdk": "1.36.0",
+    "microsoft-cognitiveservices-speech-sdk": "1.38.0",
     "openai": "^4.25.0",
     "undici": "^6.4.0"
   },

test/aws.js

@@ -19,12 +19,20 @@ test('AWS - create and cache auth token', async(t) => {
       return;
   }
   try {
-    let obj = await getAwsAuthToken(process.env.AWS_ACCESS_KEY_ID, process.env.AWS_SECRET_ACCESS_KEY, process.env.AWS_REGION);
+    let obj = await getAwsAuthToken({
+      accessKeyId: process.env.AWS_ACCESS_KEY_ID, 
+      secretAccessKey: process.env.AWS_SECRET_ACCESS_KEY,
+      region: process.env.AWS_REGION
+    });
     //console.log({obj}, 'received auth token from AWS');
     t.ok(obj.securityToken && !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);
+    obj = await getAwsAuthToken({
+      accessKeyId: process.env.AWS_ACCESS_KEY_ID, 
+      secretAccessKey: process.env.AWS_SECRET_ACCESS_KEY,
+      region: process.env.AWS_REGION
+    });
     //console.log({obj}, 'received auth token from AWS - second request');
     t.ok(obj.securityToken && obj.servedFromCache, 'successfully received access token from cache');