support verbio speech (#323)

* support verbio speech

* wip

* update speech version

* update verb specification

lib/utils/speech-utils.js

@@ -18,6 +18,7 @@ const TtsNvidiaLanguagesVoices = require('./speech-data/tts-nvidia');
 const TtsElevenlabsLanguagesVoices = require('./speech-data/tts-elevenlabs');
 const TtsWhisperLanguagesVoices = require('./speech-data/tts-whisper');
 const TtsPlayHtLanguagesVoices = require('./speech-data/tts-playht');
+const TtsVerbioLanguagesVoices = require('./speech-data/tts-verbio');
 
 const TtsModelDeepgram = require('./speech-data/tts-model-deepgram');
 const TtsModelElevenLabs = require('./speech-data/tts-model-elevenlabs');
@@ -35,6 +36,7 @@ const SttNvidiaLanguagesVoices = require('./speech-data/stt-nvidia');
 const SttCobaltLanguagesVoices = require('./speech-data/stt-cobalt');
 const SttSonioxLanguagesVoices = require('./speech-data/stt-soniox');
 const SttAssemblyaiLanguagesVoices = require('./speech-data/stt-assemblyai');
+const SttVerbioLanguagesVoices = require('./speech-data/stt-verbio');
 
 const testSonioxStt = async(logger, credentials) => {
   const api_key = credentials;
@@ -366,6 +368,43 @@ const testWellSaidStt = async(logger, credentials) => {
   return true;
 };
 
+const testVerbioTts = async(logger, synthAudio, credentials) => {
+  try {
+    await synthAudio(
+      {
+        increment: () => {},
+        histogram: () => {}
+      },
+      {
+        vendor: 'verbio',
+        credentials,
+        language: 'en-US',
+        voice: 'tommy_en-us',
+        text: 'Hi there and welcome to jambones!'
+      }
+    );
+  } catch (err) {
+    logger.info({err}, 'synth Verbio returned error');
+    throw err;
+  }
+};
+const testVerbioStt = async(logger, getVerbioAccessToken, credentials) => {
+  const token = await getVerbioAccessToken(credentials);
+  try {
+    const post = bent('https://us.rest.speechcenter.verbio.com', 'POST', 'json', {
+      'Authorization': `Bearer ${token.access_token}`,
+      'User-Agent': 'jambonz',
+      'Content-Type': 'audio/wav'
+    });
+    const json = await post('/api/v1/recognize?language=en-US&version=V1',
+      fs.readFileSync(`${__dirname}/../../data/test_audio.wav`));
+    logger.debug({json}, 'successfully speech to text from verbio');
+  } catch (err) {
+    logger.info({err}, 'testWellSaidTts returned error');
+    throw err;
+  }
+};
+
 const testAssemblyStt = async(logger, credentials) => {
   const {api_key} = credentials;
 
@@ -512,6 +551,11 @@ function decryptCredential(obj, credential, logger, isObscureKey = true) {
     const o = JSON.parse(decrypt(credential));
     obj.api_key = isObscureKey ? obscureKey(o.api_key) : o.api_key;
     obj.model_id = o.model_id;
+  } else if ('verbio' === obj.vendor) {
+    const o = JSON.parse(decrypt(credential));
+    obj.client_id = o.client_id;
+    obj.client_secret = isObscureKey ? obscureKey(o.client_secret) : o.client_secret;
+    obj.engine_version = o.engine_version;
   }
 }
 
@@ -568,6 +612,8 @@ async function getLanguagesAndVoicesForVendor(logger, vendor, credential, getTts
       return await getLanguagesVoicesForAssemblyAI(credential, getTtsVoices, logger);
     case 'whisper':
       return await getLanguagesVoicesForWhisper(credential, getTtsVoices, logger);
+    case 'verbio':
+      return await getLanguagesVoicesForVerbio(credential, getTtsVoices, logger);
     default:
       logger.info(`invalid vendor ${vendor}, return empty result`);
       throw new Error(`Invalid vendor ${vendor}`);
@@ -816,6 +862,23 @@ async function getLanguagesVoicesForWhisper(credential) {
   return tranform(TtsWhisperLanguagesVoices, undefined, TtsModelWhisper);
 }
 
+async function getLanguagesVoicesForVerbio(credentials, getTtsVoices, logger) {
+  const stt = SttVerbioLanguagesVoices.reduce((acc, v) => {
+    if (!v.version || credentials.engine_version === v.version) {
+      acc.push(v);
+    }
+    return acc;
+  }, []);
+  try {
+    const data = await getTtsVoices({vendor: 'verbio', credentials});
+    const voices = parseVerbioLanguagesVoices(data);
+    return tranform(voices, stt, undefined);
+  } catch (err) {
+    logger.info({err}, 'there is error while fetching verbio speech voices');
+    return tranform(TtsVerbioLanguagesVoices, stt, undefined);
+  }
+}
+
 function tranform(tts, stt, models) {
   return {
     ...(tts && {tts}),
@@ -943,6 +1006,29 @@ function parseMicrosoftLanguagesVoices(data) {
   }, []);
 }
 
+function parseVerbioLanguagesVoices(data) {
+  return data.reduce((acc, voice) => {
+    const languageCode = voice.language;
+    const existingLanguage = acc.find((lang) => lang.value === languageCode);
+    if (existingLanguage) {
+      existingLanguage.voices.push({
+        value: voice.voice_id,
+        name: voice.name,
+      });
+    } else {
+      acc.push({
+        value: voice.language,
+        name: voice.language,
+        voices: [{
+          value: voice.voice_id,
+          name: voice.name,
+        }]
+      });
+    }
+    return acc;
+  }, []);
+}
+
 module.exports = {
   testGoogleTts,
   testGoogleStt,
@@ -966,5 +1052,7 @@ module.exports = {
   getSpeechCredential,
   decryptCredential,
   testWhisper,
+  testVerbioTts,
+  testVerbioStt,
   getLanguagesAndVoicesForVendor
 };