From 9bbfc8ad053bbe3ed9cffadd41f95dfa8b771c3e Mon Sep 17 00:00:00 2001
From: Moishe Lettvin <moishel@gmail.com>
Date: Fri, 12 Jan 2024 11:40:49 -0500
Subject: [PATCH] Updating 06- sample

---
 .../services/daily_transport_service.py       | 38 ++++++++++---------
 .../06-listen-and-respond.py                  | 36 +++++++++---------
 2 files changed, 38 insertions(+), 36 deletions(-)

diff --git a/src/dailyai/services/daily_transport_service.py b/src/dailyai/services/daily_transport_service.py
index e26bd8195..1f1fe483b 100644
--- a/src/dailyai/services/daily_transport_service.py
+++ b/src/dailyai/services/daily_transport_service.py
@@ -50,6 +50,18 @@ class DailyTransportService(EventHandler):
         self.camera_thread = None
         self.frame_consumer_thread = None
 
+        self.transcription_settings = {
+            "language": "en",
+            "tier": "nova",
+            "model": "2-conversationalai",
+            "profanity_filter": True,
+            "redact": False,
+            "extra": {
+                "endpointing": True,
+                "punctuate": False,
+            },
+        }
+
         # This queue is used to marshal frames from the async output queue to the sync output queue
         # We need this to maintain the asynchronous behavior of asyncio queues -- to give async functions
         # a chance to run while waiting for queue items -- but also to maintain thread safety for the
@@ -170,26 +182,14 @@ class DailyTransportService(EventHandler):
         )
 
         if self.token:
-            self.transcription_queue = Queue()
-            self.client.start_transcription(
-                {
-                    "language": "en",
-                    "tier": "nova",
-                    "model": "2-conversationalai",
-                    "profanity_filter": True,
-                    "redact": False,
-                    "extra": {
-                        "endpointing": True,
-                        "punctuate": False,
-                    },
-                }
-            )
+            self.transcription_queue = asyncio.Queue()
+            self.client.start_transcription(self.transcription_settings)
 
         self.my_participant_id = self.client.participants()["local"]["id"]
 
-    def get_transcriptions(self):
+    async def get_transcriptions(self):
         while True:
-            transcript = self.transcription_queue.get()
+            transcript = await self.transcription_queue.get()
             yield transcript
 
     def get_async_output_queue(self):
@@ -259,8 +259,10 @@ class DailyTransportService(EventHandler):
         pass
 
     def on_transcription_message(self, message):
-        self.transcription_queue.put(message["text"])
-        pass
+        print("got transcription", message)
+        if self.loop:
+            asyncio.run_coroutine_threadsafe(self.transcription_queue.put(message["text"]), self.loop)
+            print("put transcription in queue", message)
 
     def on_transcription_stopped(self, stopped_by, stopped_by_error):
         pass
diff --git a/src/samples/theoretical-to-real/06-listen-and-respond.py b/src/samples/theoretical-to-real/06-listen-and-respond.py
index c36d70107..034e608f0 100644
--- a/src/samples/theoretical-to-real/06-listen-and-respond.py
+++ b/src/samples/theoretical-to-real/06-listen-and-respond.py
@@ -26,28 +26,28 @@ async def main(room_url:str, token):
     llm = AzureLLMService()
     tts = AzureTTSService()
 
-    transcribed_message = ""
-    transcription_timeout = None
+    async def handle_transcriptions():
+        messages = [
+            {"role": "system", "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio. Respond to what the user said in a creative and helpful way."},
+        ]
 
-    @transport.event_handler("on_participant_joined")
-    async def on_joined(transport, participant):
-        if participant["id"] == transport.my_participant_id:
-            return
+        sentence = ""
+        async for message in transport.get_transcriptions():
+            sentence += message
+            if sentence.endswith((".", "?", "!")):
+                messages.append({"role": "user", "content": sentence})
+                sentence = ''
 
-        async for audio_chunk in tts.run_tts("If you say something, I will respond."):
-            transport.output_queue.put(QueueFrame(FrameType.AUDIO_FRAME, audio_chunk))
+                full_response = ""
+                async for response in llm.run_llm_async_sentences(messages):
+                    full_response += response
+                    async for audio in tts.run_tts(response):
+                        transport.output_queue.put(QueueFrame(FrameType.AUDIO_FRAME, audio))
 
-    @transport.event_handler("on_transcription_message")
-    async def on_transcription_message(transport, message) -> None:
-        nonlocal transcribed_message
-        nonlocal transcription_timeout
-        print(message)
-        if message["session_id"] != transport.my_participant_id:
-            transcribed_message += message['text']
+                messages.append({"role": "assistant", "content": full_response})
 
-            print("message received", transcribed_message)
-
-    await transport.run()
+    transport.transcription_settings["extra"]["punctuate"] = True
+    await asyncio.gather(transport.run(), handle_transcriptions())
 
 
 if __name__ == "__main__":