groq worqs

pyproject.toml

@@ -12,6 +12,7 @@ dependencies = [
     "daily-python",
     "fal",
     "faster_whisper",
+    "groq",
     "google-cloud-texttospeech",
     "numpy",
     "openai",

src/dailyai/services/ai_services.py

@@ -150,7 +150,6 @@ class TTSService(AIService):
                 print(f"audio chunk size: {len(audio_chunk)}")
                 size = 8000
                 for i in range(0, len(audio_chunk), size):
-                    print(f"i is {i}")
                     yield AudioQueueFrame(audio_chunk[i : i+size])
             yield TTSCompletedFrame(text)
 

src/dailyai/services/base_transport_service.py

@@ -392,13 +392,9 @@ class BaseTransportService():
 
                     # if interrupted, we just pull frames off the queue and discard them
                     if not self._is_interrupted.is_set():
-                        print(
-                            f"~~~ not interrupted so popping frame of type {type(frame)}")
                         if frame:
                             if isinstance(frame, AudioQueueFrame):
                                 chunk = frame.data
-                                print(
-                                    f"~~~ length of this chunk: {len(chunk)}")
                                 all_audio_frames.extend(chunk)
 
                                 b.extend(chunk)

src/dailyai/services/groq_ai_services.py

@@ -1,7 +1,33 @@
-import aiohttp
+import os
+import groq
+from groq import AsyncGroq
 from dailyai.services.ai_services import LLMService
+from collections.abc import AsyncGenerator
 
 
 class GroqLLMService(LLMService):
-    def __init__(self, *, api_key, model="llama2-70b-4096"):
-        pass
+    def __init__(self, *, api_key, model="mixtral-8x7b-32768", context):
+        super().__init__(context)
+        self._model = model
+        # os.environ["GROQ_SECRET_ACCESS_KEY"] = api_key
+        
+        self._client = AsyncGroq()
+
+    async def run_llm_async(self, messages) -> AsyncGenerator[str, None]:
+        print(f"messages are {messages}")
+        try:
+            resp = await self._client.chat.completions.create(messages=messages, model=self._model)
+            print(f"got chunks from groq: {resp}")
+
+            if resp.choices[0].message.content:
+                yield resp.choices[0].message.content
+        except groq.APIConnectionError as e:
+            print("The server could not be reached")
+            print(e.__cause__)  # an underlying Exception, likely raised within httpx.
+        except groq.RateLimitError as e:
+            print("A 429 status code was received; we should back off a bit.")
+        except groq.APIStatusError as e:
+            print("Another non-200-range status code was received")
+            print(e.status_code)
+            print(e.response)
+    

src/examples/foundational/06c-listen-respond-interruptible-refactor.py

@@ -9,6 +9,7 @@ from dailyai.services.azure_ai_services import AzureLLMService, AzureTTSService
 from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
 from dailyai.services.open_ai_services import OpenAILLMService
 from dailyai.services.ai_services import FrameLogger
+from dailyai.services.groq_ai_services import GroqLLMService
 
 from examples.foundational.support.runner import configure
 
@@ -42,7 +43,7 @@ async def main(room_url: str, token):
         #     context=context)
         llm = OpenAILLMService(
             context=context, api_key=os.getenv("OPENAI_CHATGPT_API_KEY"))
-
+        llm = GroqLLMService(api_key=os.getenv("GROQ_API_KEY"), context=context)
         tts = AzureTTSService(
             api_key=os.getenv("AZURE_SPEECH_API_KEY"),
             region=os.getenv("AZURE_SPEECH_REGION"))