From 5d6d674ff61c5155d6104ec101a8151af49ae8b5 Mon Sep 17 00:00:00 2001
From: Kwindla Hultman Kramer <kwindla@gmail.com>
Date: Sun, 25 Feb 2024 21:51:08 -0800
Subject: [PATCH] some more changes

---
 src/dailyai/services/ai_services.py | 20 +++++++++++++++-----
 src/khk-hackathon/06d-listen.py     | 18 ------------------
 2 files changed, 15 insertions(+), 23 deletions(-)

diff --git a/src/dailyai/services/ai_services.py b/src/dailyai/services/ai_services.py
index 386b1b951..fd7d41b26 100644
--- a/src/dailyai/services/ai_services.py
+++ b/src/dailyai/services/ai_services.py
@@ -6,6 +6,7 @@ import datetime
 import wave
 
 from dailyai.queue_frame import (
+    QueueFrame,
     AudioQueueFrame,
     ControlQueueFrame,
     EndStreamQueueFrame,
@@ -101,8 +102,16 @@ class LLMService(AIService):
         if isinstance(frame, UserStoppedSpeakingFrame):
             print(
                 f"### Got a user stopped speaking frame, context is {self._context}")
-            async for text_chunk in self.run_llm_async(self._context):
-                yield TextQueueFrame(text_chunk)
+            async for chunk in self.run_llm_async(self._context):
+                # if we get a string, wrap it in a frame
+                if isinstance(chunk, str):
+                    yield TextQueueFrame(chunk)
+                # if we get a frame, pass it through
+                elif isinstance(chunk, QueueFrame):
+                    print(f"### Got a frame chunk: {chunk}")
+                    yield chunk
+                else:
+                    print(f"### Got an unknown chunk: {chunk}")
             yield LLMResponseEndQueueFrame()
         else:
             yield frame
@@ -149,13 +158,14 @@ class TTSService(AIService):
             async for audio_chunk in self.run_tts(text):
                 size = 8000
                 for i in range(0, len(audio_chunk), size):
-                    yield AudioQueueFrame(audio_chunk[i : i+size])
-            yield TTSCompletedFrame(text)
+                    yield AudioQueueFrame(audio_chunk[i: i+size])
+            print("### ABOUT TO YIELD TTS COMPLETED FRAME", frame)
+            yield TTSCompletedFrame(text, hasattr(frame, 'outOfBand') and frame.outOfBand)
 
     async def finalize(self):
         if self.current_sentence:
             async for audio_chunk in self.run_tts(self.current_sentence):
-               yield AudioQueueFrame(audio_chunk)
+                yield AudioQueueFrame(audio_chunk)
 
     # Convenience function to send the audio for a sentence to the given queue
     async def say(self, sentence, queue: asyncio.Queue):
diff --git a/src/khk-hackathon/06d-listen.py b/src/khk-hackathon/06d-listen.py
index 2ab01ad50..75d2d3dbe 100644
--- a/src/khk-hackathon/06d-listen.py
+++ b/src/khk-hackathon/06d-listen.py
@@ -37,20 +37,6 @@ Do not include the image model prompt in your response. The prompt must be passe
 as a parameter. 
 """
 
-do_not_respond_function = {
-    "name": "do_not_respond",
-    "description": "Call this function when the users are not talking to the robot.",
-    "parameters": {
-        "type": "object",
-        "properties": {
-            "transcribed_text": {
-                "type": "string",
-                "description": "The transcribed text from the users."
-            }
-        }
-    }
-}
-
 change_appearance_function = {
     "name": "change_appearance",
     "description": "Call this function when the users want you to change your appearance.",
@@ -66,10 +52,6 @@ change_appearance_function = {
 }
 
 tools = [
-    {
-        "type": "function",
-        "function": do_not_respond_function
-    },
     {
         "type": "function",
         "function": change_appearance_function