support mod_cartesia_transcribe (#1245)

lib/session/call-session.js

@@ -1097,6 +1097,7 @@ class CallSession extends Emitter {
           return {
             api_key: credential.api_key,
             model_id: credential.model_id,
+            stt_model_id: credential.stt_model_id,
             embedding: credential.embedding,
             options: credential.options
           };

lib/tasks/gather.js

@@ -12,6 +12,7 @@ const {
   JambonzTranscriptionEvents,
   AssemblyAiTranscriptionEvents,
   VoxistTranscriptionEvents,
+  CartesiaTranscriptionEvents,
   OpenAITranscriptionEvents,
   VadDetection,
   VerbioTranscriptionEvents,
@@ -546,6 +547,17 @@ class TaskGather extends SttTask {
           this._onVendorConnectFailure.bind(this, cs, ep));
         break;
 
+      case 'cartesia':
+        this.bugname = `${this.bugname_prefix}cartesia_transcribe`;
+        this.addCustomEventListener(ep, CartesiaTranscriptionEvents.Transcription,
+          this._onTranscription.bind(this, cs, ep));
+        this.addCustomEventListener(
+          ep, CartesiaTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
+        this.addCustomEventListener(ep, CartesiaTranscriptionEvents.Error, this._onVendorError.bind(this, cs, ep));
+        this.addCustomEventListener(ep, CartesiaTranscriptionEvents.ConnectFailure,
+          this._onVendorConnectFailure.bind(this, cs, ep));
+        break;
+
       case 'speechmatics':
         this.bugname = `${this.bugname_prefix}speechmatics_transcribe`;
         this.addCustomEventListener(

lib/tasks/transcribe.js

@@ -14,6 +14,7 @@ const {
   TranscribeStatus,
   AssemblyAiTranscriptionEvents,
   VoxistTranscriptionEvents,
+  CartesiaTranscriptionEvents,
   OpenAITranscriptionEvents,
   VerbioTranscriptionEvents,
   SpeechmaticsTranscriptionEvents
@@ -312,6 +313,17 @@ class TaskTranscribe extends SttTask {
           this._onVendorConnectFailure.bind(this, cs, ep, channel));
         break;
 
+      case 'cartesia':
+        this.bugname = `${this.bugname_prefix}cartesia_transcribe`;
+        this.addCustomEventListener(ep, CartesiaTranscriptionEvents.Transcription,
+          this._onTranscription.bind(this, cs, ep, channel));
+        this.addCustomEventListener(ep,
+          CartesiaTranscriptionEvents.Connect, this._onVendorConnect.bind(this, cs, ep));
+        this.addCustomEventListener(ep, CartesiaTranscriptionEvents.Error, this._onVendorError.bind(this, cs, ep));
+        this.addCustomEventListener(ep, CartesiaTranscriptionEvents.ConnectFailure,
+          this._onVendorConnectFailure.bind(this, cs, ep, channel));
+        break;
+
       case 'speechmatics':
         this.bugname = `${this.bugname_prefix}speechmatics_transcribe`;
         this.addCustomEventListener(

lib/utils/constants.json

@@ -167,6 +167,12 @@
     "ConnectFailure": "voxist_transcribe::connect_failed",
     "Connect": "voxist_transcribe::connect"
   },
+  "CartesiaTranscriptionEvents": {
+    "Transcription": "cartesia_transcribe::transcription",
+    "Error": "cartesia_transcribe::error",
+    "ConnectFailure": "cartesia_transcribe::connect_failed",
+    "Connect": "cartesia_transcribe::connect"
+  },
   "VadDetection": {
     "Detection": "vad_detect:detection"
   },

lib/utils/db-utils.js

@@ -110,6 +110,7 @@ const speechMapper = (cred) => {
       const o = JSON.parse(decrypt(credential));
       obj.api_key = o.api_key;
       obj.model_id = o.model_id;
+      obj.stt_model_id = o.stt_model_id;
       obj.embedding = o.embedding;
       obj.options = o.options;
     }

lib/utils/transcription-utils.js

@@ -110,6 +110,10 @@ const stickyVars = {
   voxist: [
     'VOXIST_API_KEY',
   ],
+  cartesia: [
+    'CARTESIA_API_KEY',
+    'CARTESIA_MODEL_ID'
+  ],
   speechmatics: [
     'SPEECHMATICS_API_KEY',
     'SPEECHMATICS_HOST',
@@ -555,6 +559,25 @@ const normalizeVoxist = (evt, channel, language) => {
   };
 };
 
+const normalizeCartesia = (evt, channel, language) => {
+  const copy = JSON.parse(JSON.stringify(evt));
+  return {
+    language_code: language,
+    channel_tag: channel,
+    is_final: evt.is_final,
+    alternatives: [
+      {
+        confidence: 1.00,
+        transcript: evt.text,
+      }
+    ],
+    vendor: {
+      name: 'cartesia',
+      evt: copy
+    }
+  };
+};
+
 const normalizeSpeechmatics = (evt, channel, language) => {
   const copy = JSON.parse(JSON.stringify(evt));
   const is_final = evt.message === 'AddTranscript';
@@ -636,6 +659,8 @@ module.exports = (logger) => {
         return normalizeAssemblyAi(evt, channel, language, shortUtterance);
       case 'voxist':
         return normalizeVoxist(evt, channel, language);
+      case 'cartesia':
+        return normalizeCartesia(evt, channel, language);
       case 'verbio':
         return normalizeVerbio(evt, channel, language);
       case 'speechmatics':
@@ -1008,6 +1033,16 @@ module.exports = (logger) => {
           {VOXIST_API_KEY: sttCredentials.api_key},
       };
     }
+    else if ('cartesia' === vendor) {
+      opts = {
+        ...opts,
+        ...(sttCredentials.api_key &&
+          {CARTESIA_API_KEY: sttCredentials.api_key}),
+        ...(sttCredentials.stt_model_id && {
+          CARTESIA_MODEL_ID: sttCredentials.stt_model_id
+        })
+      };
+    }
     else if ('openai' === vendor) {
       const {openaiOptions = {}} = rOpts;
       const model = openaiOptions.model || rOpts.model || sttCredentials.model_id || 'whisper-1';