support custom tts streaming (#1023)

* support custom tts streaming

* wip

* wip

* wip

* wip

* wip

* wip

* fix review comments

lib/session/call-session.js

@@ -1109,7 +1109,8 @@ class CallSession extends Emitter {
             speech_credential_sid: credential.speech_credential_sid,
             auth_token: credential.auth_token,
             custom_stt_url: credential.custom_stt_url,
-            custom_tts_url: credential.custom_tts_url
+            custom_tts_url: credential.custom_tts_url,
+            custom_tts_streaming_url: credential.custom_tts_streaming_url
           };
         }
       }

lib/tasks/tts-task.js

@@ -59,7 +59,7 @@ class TtsTask extends Task {
   }
 
   async setTtsStreamingChannelVars(vendor, language, voice, credentials, ep) {
-    const {api_key, model_id} = credentials;
+    const {api_key, model_id, custom_tts_streaming_url, auth_token} = credentials;
     const {stability, similarity_boost, use_speaker_boost, style} = this.options;
     let obj;
 
@@ -95,7 +95,18 @@ class TtsTask extends Task {
         };
         break;
       default:
-        throw new Error(`vendor ${vendor} is not supported for tts streaming yet`);
+        if (vendor.startsWith('custom:')) {
+          const use_tls = custom_tts_streaming_url.startsWith('wss://');
+          obj = {
+            CUSTOM_TTS_STREAMING_HOST: custom_tts_streaming_url.replace(/^(ws|wss):\/\//, ''),
+            CUSTOM_TTS_STREAMING_API_KEY: auth_token,
+            CUSTOM_TTS_STREAMING_VOICE_ID: voice,
+            CUSTOM_TTS_STREAMING_LANGUAGE: language || 'en',
+            CUSTOM_TTS_STREAMING_USE_TLS: use_tls
+          };
+        } else {
+          throw new Error(`vendor ${vendor} is not supported for tts streaming yet`);
+        }
     }
     this.logger.info({vendor, credentials, obj}, 'setTtsStreamingChannelVars');
 

lib/utils/constants.json

@@ -258,6 +258,11 @@
     "ConnectFailure": "elevenlabs_tts_streaming::connect_failed",
     "Connect": "elevenlabs_tts_streaming::connect"
   },
+  "CustomTtsStreamingEvents": {
+    "Empty": "custom_tts_streaming::empty",
+    "ConnectFailure": "custom_tts_streaming::connect_failed",
+    "Connect": "custom_tts_streaming::connect"
+  },
   "TtsStreamingEvents": {
     "Empty": "tts_streaming::empty",
     "Pause": "tts_streaming::pause",

lib/utils/db-utils.js

@@ -143,6 +143,7 @@ const speechMapper = (cred) => {
       obj.auth_token = o.auth_token;
       obj.custom_stt_url = o.custom_stt_url;
       obj.custom_tts_url = o.custom_tts_url;
+      obj.custom_tts_streaming_url = o.custom_tts_streaming_url;
     }
   } catch (err) {
     console.log(err);

lib/utils/tts-streaming-buffer.js

@@ -214,7 +214,7 @@ class TtsStreamingBuffer extends Emitter {
   }
 
   async _api(ep, args) {
-    const apiCmd = `uuid_${this.vendor}_tts_streaming`;
+    const apiCmd = `uuid_${this.vendor.startsWith('custom:') ? 'custom' : this.vendor}_tts_streaming`;
     const res = await ep.api(apiCmd, `^^|${args.join('|')}`);
     if (!res.body?.startsWith('+OK')) {
       this.logger.info({args}, `Error calling ${apiCmd}: ${res.body}`);
@@ -277,7 +277,8 @@ class TtsStreamingBuffer extends Emitter {
       // DH: add other vendors here as modules are added
       'deepgram',
       'cartesia',
-      'elevenlabs'
+      'elevenlabs',
+      'custom'
     ].forEach((vendor) => {
       const eventClassName = `${vendor.charAt(0).toUpperCase() + vendor.slice(1)}TtsStreamingEvents`;
       const eventClass = require('../utils/constants')[eventClassName];