Merge pull request #280 from bolna-ai/plivo-v1

adding plivo clear and mark events
voxos-ai · Jun 23, 2024 · 8415586 · 8415586
2 parents 53cd006 + 645e7ee
commit 8415586
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 8 deletions.
diff --git a/bolna/agent_manager/task_manager.py b/bolna/agent_manager/task_manager.py
@@ -1451,7 +1451,7 @@ async def __check_for_completion(self):
                     meta_info={'io': 'default', "request_id": str(uuid.uuid4()), "cached": False, "sequence_id": -1, 'format': 'wav'}
                     await self._synthesize(create_ws_data_packet("Hey, are you still there?", meta_info= meta_info))
                 else:
-                    meta_info={'io': 'twilio', "request_id": str(uuid.uuid4()), "cached": False, "sequence_id": -1, 'format': 'pcm'}
+                    meta_info={'io': self.tools["output"].get_provider(), "request_id": str(uuid.uuid4()), "cached": False, "sequence_id": -1, 'format': 'pcm'}
                     await self._synthesize(create_ws_data_packet("Hey, are you still there?", meta_info= meta_info))
 
                 #Just in case we need to clear messages sent before 
@@ -1510,7 +1510,7 @@ async def __start_transmitting_ambient_noise(self):
                 meta_info={'io': 'default', 'message_category': 'ambient_noise', "request_id": str(uuid.uuid4()), "sequence_id": -1, "type":'audio', 'format': 'wav'}
             else:
 
-                meta_info={'io': 'twilio', 'message_category': 'ambient_noise', 'stream_sid': self.stream_sid , "request_id": str(uuid.uuid4()), "cached": True, "type":'audio', "sequence_id": -1, 'format': 'pcm'}
+                meta_info={'io': self.tools["output"].get_provider(), 'message_category': 'ambient_noise', 'stream_sid': self.stream_sid , "request_id": str(uuid.uuid4()), "cached": True, "type":'audio', "sequence_id": -1, 'format': 'pcm'}
             while True:
                 logger.info(f"Before yielding ambient noise")
                 for chunk in yield_chunks_from_memory(audio, self.output_chunk_size*2 ):

diff --git a/bolna/output_handlers/telephony_providers/plivo.py b/bolna/output_handlers/telephony_providers/plivo.py
@@ -13,7 +13,16 @@ def __init__(self, websocket=None, mark_set=None, log_dir_name=None):
         io_provider = 'plivo'
 
         super().__init__(io_provider, websocket, mark_set, log_dir_name)
-        self.is_chunking_supported = False
+        self.is_chunking_supported = True
+
+    async def handle_interruption(self):
+        logger.info("interrupting because user spoke in between")
+        message_clear = {
+            "event": "clearAudio",
+            "streamId": self.stream_sid,
+        }
+        await self.websocket.send_text(json.dumps(message_clear))
+        self.mark_set = set()
 
     async def form_media_message(self, audio_data, audio_format):
         base64_audio = base64.b64encode(audio_data).decode("utf-8")
@@ -30,11 +39,9 @@ async def form_media_message(self, audio_data, audio_format):
 
     async def form_mark_message(self, mark_id):
         mark_message = {
-            "event": "mark",
-            "streamSid": self.stream_sid,
-            "mark": {
-                "name": mark_id
-            }
+            "event": "checkpoint",
+            "streamId": self.stream_sid,
+            "name": mark_id
         }
 
         return mark_message
diff --git a/bolna/synthesizer/elevenlabs_synthesizer.py b/bolna/synthesizer/elevenlabs_synthesizer.py
@@ -195,6 +195,7 @@ async def generate(self):
                     message = await self.internal_queue.get()
                     logger.info(f"Generating TTS response for message: {message}, using mulaw {self.use_mulaw}")
                     meta_info, text = message.get("meta_info"), message.get("data")
+                    audio = None
                     if self.caching:
                         if self.cache.get(text):
                             logger.info(f"Cache hit and hence returning quickly {text}")