Handle audio transcript from gpt-4o-audio and clean up logs

src/pipecat/services/anthropic/llm.py

@@ -166,7 +166,7 @@ class AnthropicLLMService(LLMService):
             await self.start_processing_metrics()
 
             logger.debug(
-                f"{self}: Generating chat [{context.system}] | [{context.get_messages_for_logging()}]"
+                f"{self}: Generating chat [{context.system}] | {context.get_messages_for_logging()}"
             )
 
             messages = context.messages

src/pipecat/services/google/llm.py

@@ -508,7 +508,7 @@ class GoogleLLMService(LLMService):
         try:
             logger.debug(
                 # f"{self}: Generating chat [{self._system_instruction}] | [{context.get_messages_for_logging()}]"
-                f"{self}: Generating chat [{context.get_messages_for_logging()}]"
+                f"{self}: Generating chat {context.get_messages_for_logging()}"
             )
 
             messages = context.messages

src/pipecat/services/openai/base_llm.py

@@ -154,7 +154,7 @@ class BaseOpenAILLMService(LLMService):
     async def _stream_chat_completions(
         self, context: OpenAILLMContext
     ) -> AsyncStream[ChatCompletionChunk]:
-        logger.debug(f"{self}: Generating chat [{context.get_messages_for_logging()}]")
+        logger.debug(f"{self}: Generating chat {context.get_messages_for_logging()}")
 
         messages: List[ChatCompletionMessageParam] = context.get_messages()
 
@@ -240,6 +240,13 @@ class BaseOpenAILLMService(LLMService):
             elif chunk.choices[0].delta.content:
                 await self.push_frame(LLMTextFrame(chunk.choices[0].delta.content))
 
+            # When gpt-4o-audio / gpt-4o-mini-audio is used for llm or stt+llm
+            # we need to get LLMTextFrame for the transcript
+            elif hasattr(chunk.choices[0].delta, "audio") and chunk.choices[0].delta.audio.get(
+                "transcript"
+            ):
+                await self.push_frame(LLMTextFrame(chunk.choices[0].delta.audio["transcript"]))
+
         # if we got a function name and arguments, check to see if it's a function with
         # a registered handler. If so, run the registered callback, save the result to
         # the context, and re-prompt to get a chat answer. If we don't have a registered