0.0.13

fixes for riva tts

lib/synth-audio.js

@@ -81,7 +81,7 @@ async function synthAudio(client, logger, stats, { account_sid,
   else if ('nvidia' === vendor) {
     assert.ok(voice, 'synthAudio requires voice when nvidia is used');
     assert.ok(language, 'synthAudio requires language when nvidia is used');
-    assert.ok(credentials.riva_uri, 'synthAudio requires riva_uri in credentials when nuance is used');
+    assert.ok(credentials.riva_server_uri, 'synthAudio requires riva_server_uri in credentials when nvidia is used');
   }
   else if ('ibm' === vendor) {
     assert.ok(voice, 'synthAudio requires voice when ibm is used');
@@ -170,8 +170,6 @@ async function synthAudio(client, logger, stats, { account_sid,
 
     client.setexAsync(key, EXPIRES, audioBuffer.toString('base64'))
       .catch((err) => logger.error(err, `error calling setex on key ${key}`));
-
-    if (['microsoft'].includes(vendor)) return {filePath, servedFromCache, rtt};
   }
 
   return new Promise((resolve, reject) => {
@@ -325,9 +323,9 @@ const synthMicrosoft = async(logger, {
               reject(cancellation.errorDetails);
               break;
             case ResultReason.SynthesizingAudioCompleted:
-              stats.increment('tts.count', ['vendor:microsoft', 'accepted:yes']);
-              resolve(result.audioData);
+              resolve(Buffer.from(result.audioData));
               synthesizer.close();
+              stats.increment('tts.count', ['vendor:microsoft', 'accepted:yes']);
               break;
             default:
               logger.info({result}, 'synthAudio: (Microsoft) unexpected result');
@@ -427,20 +425,25 @@ const synthNuance = async(client, logger, {credentials, stats, voice, model, tex
 };
 
 const synthNvidia = async(client, logger, {credentials, stats, language,  voice, model, text}) => {
-  const {riva_uri} = credentials;
-  const rivaClient = await createRivaClient(riva_uri);
-
-  const request = new SynthesizeSpeechRequest();
-  request.setVoiceName(voice);
-  request.setLanguageCode(language);
-  request.setSampleRateHz(8000);
-  request.setEncoding(AudioEncoding.LINEAR_PCM);
-  request.setText(text);
+  const {riva_server_uri} = credentials;
+  let rivaClient, request;
+  try {
+    rivaClient = await createRivaClient(riva_server_uri);
+    request = new SynthesizeSpeechRequest();
+    request.setVoiceName(voice);
+    request.setLanguageCode(language);
+    request.setSampleRateHz(8000);
+    request.setEncoding(AudioEncoding.LINEAR_PCM);
+    request.setText(text);
+  } catch (err) {
+    logger.info({err}, 'error creating riva client');
+    return Promise.reject(err);
+  }
 
   return new Promise((resolve, reject) => {
     rivaClient.synthesize(request, (err, response) => {
       if (err) {
-        console.error(err);
+        logger.info({err, voice, language}, 'error synthesizing speech using Nvidia');
         return reject(err);
       }
       resolve(Buffer.from(response.getAudio()));

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.0.10",
+  "version": "0.0.13",
   "lockfileVersion": 2,
   "requires": true,
   "packages": {
     "": {
       "name": "@jambonz/speech-utils",
-      "version": "0.0.10",
+      "version": "0.0.13",
       "license": "MIT",
       "dependencies": {
         "@aws-sdk/client-polly": "^3.303.0",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jambonz/speech-utils",
-  "version": "0.0.10",
+  "version": "0.0.13",
   "description": "TTS-related speech utilities for jambonz",
   "main": "index.js",
   "author": "Dave Horton",

test/synth.js

@@ -300,7 +300,7 @@ test('Nvidia speech synth tests', async(t) => {
     let opts = await synthAudio(stats, {
       vendor: 'nvidia',
       credentials: {
-        riva_uri: process.env.RIVA_URI,
+        riva_server_uri: process.env.RIVA_URI,
       },
       language: 'en-US',
       voice: 'English-US.Female-1',
@@ -311,7 +311,7 @@ test('Nvidia speech synth tests', async(t) => {
     opts = await synthAudio(stats, {
       vendor: 'nvidia',
       credentials: {
-        riva_uri: process.env.RIVA_URI,
+        riva_server_uri: process.env.RIVA_URI,
       },
       language: 'en-US',
       voice: 'English-US.Female-1',