initial changes for soniox (#270)

* initial changes for soniox

* changes to gather for soniox

* parse soniox stt results

* handle <end> token for soniox

* soniox: handle empty array of words

* support for soniox hints

* add soniox storage options

* update to verb specs

* add support for transcribe

* compile soniox transcripts

* gather: kill no input timer for soniox when we get interim results

* fix buffering of soniox transcripts

* fix for compiling soniox transcript

* another fix for compiling soniox transcript

* another fix

* handling of <end> token

* fix soniox bug

* gather: fixes for soniox continous asr

* fix undefined variable reference

* fix prev commit

* bugfix: allow verb_status requests

* gather: for soniox no need to restart transcription after final transcription received

* update verb specs

* update verb specs, fixes for continuous asr:

lib/session/call-session.js

@@ -645,6 +645,12 @@ class CallSession extends Emitter {
             api_key: credential.api_key
           };
         }
+        else if ('soniox' === vendor) {
+          return {
+            speech_credential_sid: credential.speech_credential_sid,
+            api_key: credential.api_key
+          };
+        }
         else if ('ibm' === vendor) {
           return {
             speech_credential_sid: credential.speech_credential_sid,

lib/tasks/gather.js

@@ -7,6 +7,7 @@ const {
   AwsTranscriptionEvents,
   AzureTranscriptionEvents,
   DeepgramTranscriptionEvents,
+  SonioxTranscriptionEvents,
   IbmTranscriptionEvents,
   NvidiaTranscriptionEvents
 } = require('../utils/constants');
@@ -33,11 +34,13 @@ class TaskGather extends Task {
       setChannelVarsForStt,
       normalizeTranscription,
       removeSpeechListeners,
-      setSpeechCredentialsAtRuntime
+      setSpeechCredentialsAtRuntime,
+      compileSonioxTranscripts
     } = require('../utils/transcription-utils')(logger);
     this.setChannelVarsForStt = setChannelVarsForStt;
     this.normalizeTranscription = normalizeTranscription;
     this.removeSpeechListeners = removeSpeechListeners;
+    this.compileSonioxTranscripts = compileSonioxTranscripts;
 
     [
       'finishOnKey', 'input', 'numDigits', 'minDigits', 'maxDigits',
@@ -85,6 +88,9 @@ class TaskGather extends Task {
     /* buffer speech for continuous asr */
     this._bufferedTranscripts = [];
 
+    /* buffer for soniox transcripts */
+    this._sonioxTranscripts = [];
+
     this.parentTask = parentTask;
   }
 
@@ -288,6 +294,7 @@ class TaskGather extends Task {
     this._killAudio(cs);
     this.ep.removeAllListeners('dtmf');
     clearTimeout(this.interDigitTimer);
+    this._clearAsrTimer();
     this.playTask?.span.end();
     this.sayTask?.span.end();
     this._resolve('killed');
@@ -389,6 +396,13 @@ class TaskGather extends Task {
           this._onDeepGramConnectFailure.bind(this, cs, ep));
         break;
 
+      case 'soniox':
+        this.bugname = 'soniox_transcribe';
+        ep.addCustomEventListener(SonioxTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
+        ep.addCustomEventListener(SonioxTranscriptionEvents.Error,
+          this._onSonioxError.bind(this, cs, ep));
+        break;
+
       case 'ibm':
         this.bugname = 'ibm_transcribe';
         ep.addCustomEventListener(IbmTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
@@ -523,7 +537,7 @@ class TaskGather extends Task {
     // make sure this is not a transcript from answering machine detection
     const bugname = fsEvent.getHeader('media-bugname');
     const finished = fsEvent.getHeader('transcription-session-finished');
-    this.logger.debug({evt, bugname, finished}, 'Gather:_onTranscription');
+    this.logger.debug({evt, bugname, finished}, `Gather:_onTranscription for vendor ${this.vendor}`);
     if (bugname && this.bugname !== bugname) return;
 
     if (this.vendor === 'ibm') {
@@ -544,9 +558,8 @@ class TaskGather extends Task {
 
     /* count words for bargein feature */
     const words = evt.alternatives[0]?.transcript.split(' ').length;
-    const bufferedWords = this._bufferedTranscripts.reduce((count, e) => {
+    const bufferedWords = this._sonioxTranscripts.length +
-      return count + e.alternatives[0]?.transcript.split(' ').length;
+      this._bufferedTranscripts.reduce((count, e) => count + e.alternatives[0]?.transcript.split(' ').length, 0);
-    }, 0);
 
     if (evt.is_final) {
       if (evt.alternatives[0].transcript === '' && !this.callSession.callGone && !this.killed) {
@@ -555,7 +568,6 @@ class TaskGather extends Task {
         }
         else {
           this.logger.info({evt}, 'TaskGather:_onTranscription - got empty transcript, continue listening');
-          //this._startTranscribing(ep);
         }
         return;
       }
@@ -579,7 +591,9 @@ class TaskGather extends Task {
           return this._resolve(this._bufferedTranscripts.length > 0 ? 'speech' : 'timeout');
         }
         this._startAsrTimer();
-        return this._startTranscribing(ep);
+
+        /* some STT engines will keep listening after a final response, so no need to restart */
+        if (!['soniox', 'aws', 'microsoft', 'deepgram'].includes(this.vendor)) this._startTranscribing(ep);
       }
       else {
         if (this.bargein && (words + bufferedWords) < this.minBargeinWordCount) {
@@ -590,6 +604,12 @@ class TaskGather extends Task {
           return;
         }
         else {
+          if (this.vendor === 'soniox') {
+            /* compile transcripts into one */
+            this._sonioxTranscripts.push(evt.vendor.finalWords);
+            evt = this.compileSonioxTranscripts(this._sonioxTranscripts, 1, this.language);
+            this._sonioxTranscripts = [];
+          }
           this._resolve('speech', evt);
         }
       }
@@ -613,6 +633,13 @@ class TaskGather extends Task {
         this.cs.requestor.request('verb:hook', this.partialResultHook,  Object.assign({speech: evt},
           this.cs.callInfo, httpHeaders));
       }
+      if (this.vendor === 'soniox') {
+        this._clearTimer();
+        if (evt.vendor.finalWords.length) {
+          this.logger.debug({evt}, 'TaskGather:_onTranscription - buffering soniox transcript');
+          this._sonioxTranscripts.push(evt.vendor.finalWords);
+        }
+      }
     }
   }
   _onEndOfUtterance(cs, ep) {
@@ -643,6 +670,9 @@ class TaskGather extends Task {
       return this._resolve('timeout');
     }
   }
+  _onSonioxError(cs, ep, evt) {
+    this.logger.info({evt}, 'TaskGather:_onSonioxError');
+  }
   _onNvidiaError(cs, ep, evt) {
     this.logger.info({evt}, 'TaskGather:_onNvidiaError');
   }

lib/tasks/transcribe.js

@@ -3,10 +3,11 @@ const {
   TaskName,
   TaskPreconditions,
   GoogleTranscriptionEvents,
-  AzureTranscriptionEvents,
-  AwsTranscriptionEvents,
   NuanceTranscriptionEvents,
+  AwsTranscriptionEvents,
+  AzureTranscriptionEvents,
   DeepgramTranscriptionEvents,
+  SonioxTranscriptionEvents,
   IbmTranscriptionEvents,
   NvidiaTranscriptionEvents
 } = require('../utils/constants');
@@ -195,7 +196,12 @@ class TaskTranscribe extends Task {
         ep.addCustomEventListener(DeepgramTranscriptionEvents.ConnectFailure,
           this._onDeepGramConnectFailure.bind(this, cs, ep, channel));
         break;
-
+      case 'soniox':
+        this.bugname = 'soniox_transcribe';
+        ep.addCustomEventListener(SonioxTranscriptionEvents.Transcription, this._onTranscription.bind(this, cs, ep));
+        ep.addCustomEventListener(SonioxTranscriptionEvents.Error,
+          this._onSonioxError.bind(this, cs, ep));
+        break;
       case 'ibm':
         this.bugname = 'ibm_transcribe';
         ep.addCustomEventListener(IbmTranscriptionEvents.Transcription,
@@ -326,8 +332,11 @@ class TaskTranscribe extends Task {
       return this._resolve('timeout');
     }
   }
+  _onSonioxError(cs, ep, evt) {
+    this.logger.info({evt}, 'TaskTranscribe:_onSonioxError');
+  }
   _onNvidiaError(cs, ep, evt) {
-    this.logger.info({evt}, 'TaskGather:_onNvidiaError');
+    this.logger.info({evt}, 'TaskTranscribe:_onNvidiaError');
   }
   _onDeepgramConnect(_cs, _ep) {
     this.logger.debug('TaskTranscribe:_onDeepgramConnect');
@@ -365,7 +374,7 @@ class TaskTranscribe extends Task {
     this.notifyTaskDone();
   }
   _onIbmError(cs, _ep, _channel, evt) {
-    this.logger.info({evt}, 'TaskGather:_onIbmError');
+    this.logger.info({evt}, 'TaskTranscribe:_onIbmError');
   }
 
 

lib/utils/constants.json

@@ -86,6 +86,10 @@
     "ConnectFailure": "deepgram_transcribe::connect_failed",
     "Connect": "deepgram_transcribe::connect"
   },
+  "SonioxTranscriptionEvents": {
+    "Transcription": "soniox_transcribe::transcription",
+    "Error": "soniox_transcribe::error"
+  },
   "IbmTranscriptionEvents": {
     "Transcription": "ibm_transcribe::transcription",
     "ConnectFailure": "ibm_transcribe::connect_failed",
@@ -147,6 +151,7 @@
     "queue:status",
     "dial:confirm",
     "verb:hook",
+    "verb:status",
     "jambonz:error"
   ],
   "RecordState": {

lib/utils/db-utils.js

@@ -62,6 +62,10 @@ const speechMapper = (cred) => {
       const o = JSON.parse(decrypt(credential));
       obj.api_key = o.api_key;
     }
+    else if ('soniox' === obj.vendor) {
+      const o = JSON.parse(decrypt(credential));
+      obj.api_key = o.api_key;
+    }
   } catch (err) {
     console.log(err);
   }
@@ -86,8 +90,10 @@ module.exports = (logger, srf) => {
     const haveWellsaid = speech.find((s) => s.vendor === 'wellsaid');
     const haveNuance = speech.find((s) => s.vendor === 'nuance');
     const haveDeepgram = speech.find((s) => s.vendor === 'deepgram');
+    const haveSoniox = speech.find((s) => s.vendor === 'soniox');
     const haveIbm = speech.find((s) => s.vendor === 'ibm');
-    if (!haveGoogle || !haveAws || !haveMicrosoft || !haveWellsaid || !haveNuance || !haveIbm || !haveDeepgram) {
+    if (!haveGoogle || !haveAws || !haveMicrosoft || !haveWellsaid ||
+      !haveNuance || !haveIbm || !haveDeepgram || !haveSoniox) {
       const [r3] = await pp.query(sqlSpeechCredentialsForSP, account_sid);
       if (r3.length) {
         if (!haveGoogle) {
@@ -114,6 +120,10 @@ module.exports = (logger, srf) => {
           const deepgram = r3.find((s) => s.vendor === 'deepgram');
           if (deepgram) speech.push(speechMapper(deepgram));
         }
+        if (!haveSoniox) {
+          const soniox = r3.find((s) => s.vendor === 'soniox');
+          if (soniox) speech.push(speechMapper(soniox));
+        }
         if (!haveIbm) {
           const ibm = r3.find((s) => s.vendor === 'ibm');
           if (ibm) speech.push(speechMapper(ibm));

lib/utils/transcription-utils.js

@@ -5,6 +5,7 @@ const {
   AwsTranscriptionEvents,
   NuanceTranscriptionEvents,
   DeepgramTranscriptionEvents,
+  SonioxTranscriptionEvents,
   NvidiaTranscriptionEvents
 } = require('./constants');
 
@@ -88,9 +89,70 @@ const stickyVars = {
   ],
   nvidia: [
     'NVIDIA_HINTS'
+  ],
+  soniox: [
+    'SONIOX_PROFANITY_FILTER',
+    'SONIOX_MODEL'
   ]
 };
 
+const compileSonioxTranscripts = (finalWordChunks, channel, language) => {
+  const words = finalWordChunks.flat();
+  const transcript = words.reduce((acc, word) => {
+    if (word.text === '<end>') return acc;
+    if ([',', '.', '?', '!'].includes(word.text)) return `${acc}${word.text}`;
+    return `${acc} ${word.text}`;
+  }, '').trim();
+  const realWords = words.filter((word) => ![',.!?;'].includes(word.text) && word.text !== '<end>');
+  const confidence = realWords.reduce((acc, word) => acc + word.confidence, 0) / realWords.length;
+  const alternatives = [{transcript, confidence}];
+  return {
+    language_code: language,
+    channel_tag: channel,
+    is_final: true,
+    alternatives,
+    vendor: {
+      name: 'soniox',
+      evt: words
+    }
+  };
+};
+
+const normalizeSoniox = (evt, channel, language) => {
+  const copy = JSON.parse(JSON.stringify(evt));
+
+  /* an <end> token indicates the end of an utterance */
+  const endTokenPos = evt.words.map((w) => w.text).indexOf('<end>');
+  const endpointReached = endTokenPos !== -1;
+  const words = endpointReached ? evt.words.slice(0, endTokenPos) : evt.words;
+
+  /* note: we can safely ignore words after the <end> token as they will be returned again */
+  const finalWords = words.filter((word) => word.is_final);
+  const nonFinalWords = words.filter((word) => !word.is_final);
+
+  const is_final = endpointReached && finalWords.length > 0;
+  const transcript = words.reduce((acc, word) => {
+    if ([',', '.', '?', '!'].includes(word.text)) return `${acc}${word.text}`;
+    else return `${acc} ${word.text}`;
+  }, '').trim();
+  const realWords = words.filter((word) => ![',.!?;'].includes(word.text) && word.text !== '<end>');
+  const confidence = realWords.reduce((acc, word) => acc + word.confidence, 0) / realWords.length;
+  const alternatives = [{transcript, confidence}];
+  return {
+    language_code: language,
+    channel_tag: channel,
+    is_final,
+    alternatives,
+    vendor: {
+      name: 'soniox',
+      endpointReached,
+      evt: copy,
+      finalWords,
+      nonFinalWords
+    }
+  };
+};
+
 const normalizeDeepgram = (evt, channel, language) => {
   const copy = JSON.parse(JSON.stringify(evt));
   const alternatives = (evt.channel?.alternatives || [])
@@ -237,6 +299,8 @@ module.exports = (logger) => {
         return normalizeIbm(evt, channel, language);
       case 'nvidia':
         return normalizeNvidia(evt, channel, language);
+      case 'soniox':
+        return normalizeSoniox(evt, channel, language);
       default:
         logger.error(`Unknown vendor ${vendor}`);
         return evt;
@@ -441,6 +505,29 @@ module.exports = (logger) => {
           {DEEPGRAM_SPEECH_TAG: deepgramOptions.tag}
       };
     }
+    else if ('soniox' === rOpts.vendor) {
+      const {sonioxOptions = {}} = rOpts;
+      const {storage = {}} = sonioxOptions;
+      opts = {
+        ...opts,
+        ...(sttCredentials.api_key) &&
+          {SONIOX_API_KEY: sttCredentials.api_key},
+        ...(rOpts.hints.length > 0 && typeof rOpts.hints[0] === 'string' &&
+          {SONIOX_HINTS: rOpts.hints.join(',')}),
+        ...(rOpts.hints.length > 0 && typeof rOpts.hints[0] === 'object' &&
+          {SONIOX_HINTS: JSON.stringify(rOpts.hints)}),
+        ...(typeof rOpts.hintsBoost === 'number' &&
+          {SONIOX_HINTS_BOOST: rOpts.hintsBoost}),
+        ...(sonioxOptions.model) &&
+          {SONIOX_MODEL: sonioxOptions.model},
+        ...((sonioxOptions.profanityFilter || rOpts.profanityFilter) && {SONIOX_PROFANITY_FILTER: 1}),
+        ...(storage?.id && {SONIOX_STORAGE_ID: storage.id}),
+        ...(storage?.id && storage?.title && {SONIOX_STORAGE_TITLE: storage.title}),
+        ...(storage?.id && storage?.disableStoreAudio && {SONIOX_STORAGE_DISABLE_AUDIO: 1}),
+        ...(storage?.id && storage?.disableStoreTranscript && {SONIOX_STORAGE_DISABLE_TRANSCRIPT: 1}),
+        ...(storage?.id && storage?.disableSearch && {SONIOX_STORAGE_DISABLE_SEARCH: 1})
+      };
+    }
     else if ('ibm' === rOpts.vendor) {
       const {ibmOptions = {}} = rOpts;
       opts = {
@@ -524,6 +611,9 @@ module.exports = (logger) => {
     ep.removeCustomEventListener(DeepgramTranscriptionEvents.Connect);
     ep.removeCustomEventListener(DeepgramTranscriptionEvents.ConnectFailure);
 
+    ep.removeCustomEventListener(SonioxTranscriptionEvents.Transcription);
+    ep.removeCustomEventListener(SonioxTranscriptionEvents.Error);
+
     ep.removeCustomEventListener(NvidiaTranscriptionEvents.Transcription);
     ep.removeCustomEventListener(NvidiaTranscriptionEvents.TranscriptionComplete);
     ep.removeCustomEventListener(NvidiaTranscriptionEvents.StartOfSpeech);
@@ -534,8 +624,9 @@ module.exports = (logger) => {
   const setSpeechCredentialsAtRuntime = (recognizer) => {
     if (!recognizer) return;
     if (recognizer.vendor === 'nuance') {
-      const {clientId, secret} = recognizer.nuanceOptions || {};
+      const {clientId, secret, kryptonEndpoint} = recognizer.nuanceOptions || {};
       if (clientId && secret) return {client_id: clientId, secret};
+      if (kryptonEndpoint) return {krypton_endpoint: kryptonEndpoint};
     }
     else if (recognizer.vendor === 'nvidia') {
       const {rivaUri} = recognizer.nvidiaOptions || {};
@@ -545,6 +636,10 @@ module.exports = (logger) => {
       const {apiKey} = recognizer.deepgramOptions || {};
       if (apiKey) return {api_key: apiKey};
     }
+    else if (recognizer.vendor === 'soniox') {
+      const {apiKey} = recognizer.sonioxOptions || {};
+      if (apiKey) return {api_key: apiKey};
+    }
     else if (recognizer.vendor === 'ibm') {
       const {ttsApiKey, ttsRegion, sttApiKey, sttRegion, instanceId} = recognizer.ibmOptions || {};
       if (ttsApiKey || sttApiKey) return {
@@ -561,6 +656,7 @@ module.exports = (logger) => {
     normalizeTranscription,
     setChannelVarsForStt,
     removeSpeechListeners,
-    setSpeechCredentialsAtRuntime
+    setSpeechCredentialsAtRuntime,
+    compileSonioxTranscripts
   };
 };

package-lock.json

@@ -14,7 +14,7 @@
         "@jambonz/realtimedb-helpers": "^0.6.5",
         "@jambonz/stats-collector": "^0.1.6",
         "@jambonz/time-series": "^0.2.5",
-        "@jambonz/verb-specifications": "^0.0.4",
+        "@jambonz/verb-specifications": "^0.0.11",
         "@opentelemetry/api": "^1.4.0",
         "@opentelemetry/exporter-jaeger": "^1.9.0",
         "@opentelemetry/exporter-trace-otlp-http": "^0.35.0",
@@ -28,7 +28,7 @@
         "bent": "^7.3.12",
         "debug": "^4.3.4",
         "deepcopy": "^2.1.0",
-        "drachtio-fsmrf": "^3.0.18",
+        "drachtio-fsmrf": "^3.0.19",
         "drachtio-srf": "^4.5.23",
         "express": "^4.18.2",
         "ip": "^1.1.8",
@@ -675,9 +675,9 @@
       }
     },
     "node_modules/@jambonz/verb-specifications": {
-      "version": "0.0.4",
+      "version": "0.0.11",
-      "resolved": "https://registry.npmjs.org/@jambonz/verb-specifications/-/verb-specifications-0.0.4.tgz",
+      "resolved": "https://registry.npmjs.org/@jambonz/verb-specifications/-/verb-specifications-0.0.11.tgz",
-      "integrity": "sha512-Y3aN6NyiTqi0AdGFG9/sZwurG2ogCYANJpekVThlUNHTtg6UVIY6Tt+ND8+Hfc6UvNRZWAQIkugzpZvJFJ0faA==",
+      "integrity": "sha512-T5ZITmOkcXOIBSH/vSGIQ1PXyZfOtdpo/DpJQ82gdIfYhm4AotMVbS5YABj1RFj4ergNnkoHmmhgJttgpMf0dw==",
       "dependencies": {
         "debug": "^4.3.4",
         "pino": "^8.8.0"
@@ -2228,9 +2228,9 @@
       }
     },
     "node_modules/drachtio-fsmrf": {
-      "version": "3.0.18",
+      "version": "3.0.19",
-      "resolved": "https://registry.npmjs.org/drachtio-fsmrf/-/drachtio-fsmrf-3.0.18.tgz",
+      "resolved": "https://registry.npmjs.org/drachtio-fsmrf/-/drachtio-fsmrf-3.0.19.tgz",
-      "integrity": "sha512-vWltwmIYeAapE2R5ise2HZgRS+8aGlQMXN6aTUxKGnvzGPD9+D29or+Y8HtodYG/2RKaPcaeUOjjsmnUaoWU2A==",
+      "integrity": "sha512-MMofkpNCMxk58mIaXOe2jwTPwZ5f9AEH0eqqBaBR5a9qP8n7/xES18Qd7KaqDAOocTn99t0cl1k8Y+54F+lwfQ==",
       "dependencies": {
         "camel-case": "^4.1.2",
         "debug": "^2.6.9",
@@ -8132,9 +8132,9 @@
       }
     },
     "@jambonz/verb-specifications": {
-      "version": "0.0.4",
+      "version": "0.0.11",
-      "resolved": "https://registry.npmjs.org/@jambonz/verb-specifications/-/verb-specifications-0.0.4.tgz",
+      "resolved": "https://registry.npmjs.org/@jambonz/verb-specifications/-/verb-specifications-0.0.11.tgz",
-      "integrity": "sha512-Y3aN6NyiTqi0AdGFG9/sZwurG2ogCYANJpekVThlUNHTtg6UVIY6Tt+ND8+Hfc6UvNRZWAQIkugzpZvJFJ0faA==",
+      "integrity": "sha512-T5ZITmOkcXOIBSH/vSGIQ1PXyZfOtdpo/DpJQ82gdIfYhm4AotMVbS5YABj1RFj4ergNnkoHmmhgJttgpMf0dw==",
       "requires": {
         "debug": "^4.3.4",
         "pino": "^8.8.0"
@@ -9324,9 +9324,9 @@
       }
     },
     "drachtio-fsmrf": {
-      "version": "3.0.18",
+      "version": "3.0.19",
-      "resolved": "https://registry.npmjs.org/drachtio-fsmrf/-/drachtio-fsmrf-3.0.18.tgz",
+      "resolved": "https://registry.npmjs.org/drachtio-fsmrf/-/drachtio-fsmrf-3.0.19.tgz",
-      "integrity": "sha512-vWltwmIYeAapE2R5ise2HZgRS+8aGlQMXN6aTUxKGnvzGPD9+D29or+Y8HtodYG/2RKaPcaeUOjjsmnUaoWU2A==",
+      "integrity": "sha512-MMofkpNCMxk58mIaXOe2jwTPwZ5f9AEH0eqqBaBR5a9qP8n7/xES18Qd7KaqDAOocTn99t0cl1k8Y+54F+lwfQ==",
       "requires": {
         "camel-case": "^4.1.2",
         "debug": "^2.6.9",

package.json

@@ -29,7 +29,7 @@
     "@jambonz/realtimedb-helpers": "^0.6.5",
     "@jambonz/stats-collector": "^0.1.6",
     "@jambonz/time-series": "^0.2.5",
-    "@jambonz/verb-specifications": "^0.0.4",
+    "@jambonz/verb-specifications": "^0.0.11",
     "@opentelemetry/api": "^1.4.0",
     "@opentelemetry/exporter-jaeger": "^1.9.0",
     "@opentelemetry/exporter-trace-otlp-http": "^0.35.0",
@@ -43,7 +43,7 @@
     "bent": "^7.3.12",
     "debug": "^4.3.4",
     "deepcopy": "^2.1.0",
-    "drachtio-fsmrf": "^3.0.18",
+    "drachtio-fsmrf": "^3.0.19",
     "drachtio-srf": "^4.5.23",
     "express": "^4.18.2",
     "ip": "^1.1.8",

test/gather-tests.js

@@ -207,6 +207,48 @@ test('\'gather\' test - deepgram', async(t) => {
     //console.log(JSON.stringify(obj));
     t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().startsWith('i\'d like to speak to customer support'),
       'gather: succeeds when using deepgram credentials');
+    disconnect();
+  } catch (err) {
+    console.log(`error received: ${err}`);
+    disconnect();
+    t.error(err);
+  }
+});
+
+test('\'gather\' test - soniox', async(t) => {
+  if (!process.env.SONIOX_API_KEY ) {
+    t.pass('skipping soniox tests');
+    return t.end();
+  }
+  clearModule.all();
+  const {srf, disconnect} = require('../app');
+
+  try {
+    await connect(srf);
+    // GIVEN
+    let verbs = [
+      {
+        "verb": "gather",
+        "input": ["speech"],
+        "recognizer": {
+          "vendor": "deepgram",
+          "hints": ["customer support", "sales", "human resources", "HR"],
+          "deepgramOptions": {
+            "apiKey": process.env.SONIOX_API_KEY
+          }
+        },
+        "timeout": 10,
+        "actionHook": "/actionHook"
+      }
+    ];
+    let from = "gather_success";
+    provisionCallHook(from, verbs);
+    // THEN
+    await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
+    let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
+    console.log(JSON.stringify(obj));
+    t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().startsWith('i\'d like to speak to customer support'),
+      'gather: succeeds when using soniox credentials');
 
     disconnect();
   } catch (err) {

test/transcribe-tests.js

@@ -143,7 +143,7 @@ test('\'transcribe\' test - deepgram', async(t) => {
       {
         "verb": "transcribe",
         "recognizer": {
-          "vendor": "aws",
+          "vendor": "deepgram",
           "hints": ["customer support", "sales", "human resources", "HR"],
           "deepgramOptions": {
             "apiKey": process.env.DEEPGRAM_API_KEY
@@ -167,3 +167,44 @@ test('\'transcribe\' test - deepgram', async(t) => {
     t.error(err);
   }
 });
+
+test('\'transcribe\' test - soniox', async(t) => {
+  if (!process.env.SONIOX_API_KEY ) {
+    t.pass('skipping soniox tests');
+    return t.end();
+  }
+  clearModule.all();
+  const {srf, disconnect} = require('../app');
+
+  try {
+    await connect(srf);
+    // GIVEN
+    let verbs = [
+      {
+        "verb": "transcribe",
+        "recognizer": {
+          "vendor": "soniox",
+          "hints": ["customer support", "sales", "human resources", "HR"],
+          "deepgramOptions": {
+            "apiKey": process.env.SONIOX_API_KEY
+          }
+        },
+        "transcriptionHook": "/transcriptionHook"
+      }
+    ];
+    let from = "gather_success";
+    provisionCallHook(from, verbs);
+    // THEN
+    await sippUac('uac-gather-account-creds-success.xml', '172.38.0.10', from);
+    let obj = await getJSON(`http://127.0.0.1:3100/lastRequest/${from}_actionHook`);
+    console.log(JSON.stringify(obj));
+    t.ok(obj.body.speech.alternatives[0].transcript.toLowerCase().startsWith('i\'d like to speak to customer support'),
+      'transcribe: succeeds when using soniox credentials');
+
+    disconnect();
+  } catch (err) {
+    console.log(`error received: ${err}`);
+    disconnect();
+    t.error(err);
+  }
+});