Some cleanup, bringing azure ai service up to date

src/dailyai/async_processor/async_processor.py

@@ -318,7 +318,7 @@ class Response(AsyncProcessor):
             self.play_thread.join()
 
 
-@dataclass
+@dataclass(frozen=True)
 class ConversationProcessorCollection:
     introduction: Optional[Type[Response]] = None
     waiting: Optional[Type[Response]] = None

src/dailyai/orchestrator.py

@@ -38,14 +38,24 @@ class OrchestratorConfig:
     bot_name: str
     expiration: float
 
+# Note that we use this as a default parameter value in the Orchestrator
+# constructor. The dataclass is defined with Frozen=True, so this should
+# be safe.
+default_conversation_collection = ConversationProcessorCollection(
+    introduction=Response,
+    waiting=None,
+    response=Response,
+    goodbye=None,
+)
 
 class Orchestrator(EventHandler):
+
     def __init__(
         self,
         daily_config: OrchestratorConfig,
         ai_service_config: AIServiceConfig,
-        conversation_processors: ConversationProcessorCollection,
         message_handler: MessageHandler,
+        conversation_processors: ConversationProcessorCollection = default_conversation_collection,
         tracer=None,
     ):
         self.bot_name: str = daily_config.bot_name
@@ -272,8 +282,6 @@ class Orchestrator(EventHandler):
         except Exception as e:
             self.logger.error(f"Exception {e} in camera thread.")
 
-        print("==== camera thread exitings")
-
     def handle_user_started_talking(self):
         # TODO: allow configuration of the timer timeout
         self.logger.error("user started talking")

src/dailyai/services/azure_ai_services.py

@@ -14,13 +14,13 @@ from PIL import Image
 from azure.cognitiveservices.speech import SpeechSynthesizer, SpeechConfig, ResultReason, CancellationReason
 
 class AzureTTSService(TTSService):
-    def __init__(self):
+    def __init__(self, speech_key=None, speech_region=None):
         super().__init__()
 
-        self.speech_key = os.getenv("AZURE_SPEECH_SERVICE_KEY")
+        speech_key = speech_key or os.getenv("AZURE_SPEECH_SERVICE_KEY")
-        self.speech_region = os.getenv("AZURE_SPEECH_SERVICE_REGION")
+        speech_region = speech_region or os.getenv("AZURE_SPEECH_SERVICE_REGION")
 
-        self.speech_config = SpeechConfig(subscription=self.speech_key, region=self.speech_region)
+        self.speech_config = SpeechConfig(subscription=speech_key, region=speech_region)
         self.speech_synthesizer = SpeechSynthesizer(speech_config=self.speech_config, audio_config=None)
 
     def run_tts(self, sentence) -> Generator[bytes, None, None]:
@@ -46,19 +46,23 @@ class AzureTTSService(TTSService):
                 self.logger.info("Error details: {}".format(cancellation_details.error_details))
 
 class AzureLLMService(LLMService):
-    def __init__(self):
+    def __init__(self, api_key=None, azure_endpoint=None, api_version=None, model=None):
         super().__init__()
+        api_key = api_key or os.getenv("AZURE_CHATGPT_KEY")
+        azure_endpoint = azure_endpoint or os.getenv("AZURE_CHATGPT_ENDPOINT")
+        api_version = api_version or "2023-12-01-preview"
         self.client = AzureOpenAI(
-            api_key=os.getenv("AZURE_CHATGPT_KEY"),
+            api_key=api_key,
-            azure_endpoint=os.getenv("AZURE_CHATGPT_ENDPOINT"),
+            azure_endpoint=azure_endpoint,
-            api_version="2023-12-01-preview",
+            api_version=api_version,
         )
+        self.model = model or os.getenv("AZURE_CHATGPT_DEPLOYMENT_ID")
 
     def get_response(self, messages, stream):
         return self.client.chat.completions.create(
             stream=stream,
             messages=messages,
-            model=os.getenv("AZURE_CHATGPT_DEPLOYMENT_ID"),
+            model=self.model,
         )
 
     def run_llm_async(self, messages) -> Generator[str, None, None]:
@@ -89,16 +93,26 @@ class AzureLLMService(LLMService):
 
 class AzureImageGenService(ImageGenService):
 
+    def __init__(self, api_key=None, azure_endpoint=None, api_version=None, model=None):
+        super().__init__()
+
+        api_key = api_key or os.getenv("AZURE_DALLE_KEY")
+        azure_endpoint = azure_endpoint or os.getenv("AZURE_DALLE_ENDPOINT")
+        api_version = api_version or "2023-12-01-preview"
+        self.model = model or os.getenv("AZURE_DALLE_DEPLOYMENT_ID")
+
+        self.client = AzureOpenAI(
+            api_key=api_key,
+            azure_endpoint=azure_endpoint,
+            api_version=api_version,
+        )
+
     def run_image_gen(self, sentence) -> tuple[str, Image.Image]:
         self.logger.info("Generating azure image", sentence)
 
-        image = OpenAI().images.generate(
+        image = self.client.images.generate(
-            api_type="azure",
+            model=self.model,
-            api_version="2023-06-01-preview",
+            prompt=sentence,
-            api_key=os.getenv("AZURE_DALLE_KEY"),
-            api_base=os.getenv("AZURE_DALLE_ENDPOINT"),
-            deployment_id=os.getenv("AZURE_DALLE_DEPLOYMENT_ID"),
-            prompt=f"{sentence} in the style of {self.image_style}",
             n=1,
             size=f"1024x1024",
         )

src/samples/simple-sample/simple-sample.py

@@ -29,15 +29,6 @@ def add_bot_to_room(room_url, token, expiration) -> None:
     """
     )
 
-    # Use the standard Response classes for the intro and response. Do nothing while waiting
-    # or saying goodbye.
-    conversation_processors = ConversationProcessorCollection(
-        introduction=Response,
-        waiting=None,
-        response=Response,
-        goodbye=None,
-    )
-
     # Use Azure services for the TTS, image generation, and LLM.
     # Note that you'll need to set the following environment variables:
     # - AZURE_SPEECH_SERVICE_KEY
@@ -66,7 +57,6 @@ def add_bot_to_room(room_url, token, expiration) -> None:
     orchestrator = Orchestrator(
         orchestrator_config,
         services,
-        conversation_processors,
         message_handler,
     )
     orchestrator.start()