chore: streaming audio

app.py

@@ -118,7 +118,7 @@ def generate_response_and_audio(audio_bytes: bytes, state: AppState):
 
         full_response = ""
         asr_result = ""
-        audios = []
+        final_audio = b""
         id = uuid.uuid4()
 
         for chunk in stream:
@@ -134,9 +134,8 @@ def generate_response_and_audio(audio_bytes: bytes, state: AppState):
                 full_response += content
                 yield id, full_response, asr_result, None, state
             if audio:
-                audios.extend(audio)
-
-        final_audio = b"".join([base64.b64decode(a) for a in audios])
+                final_audio = b"".join([base64.b64decode(a) for a in audio])
+                yield id, full_response, asr_result, final_audio, state
 
         yield id, full_response, asr_result, final_audio, state
 
@@ -243,7 +242,7 @@ with gr.Blocks() as demo:
             input_audio = gr.Audio(
                 label="Input Audio", sources="microphone", type="numpy"
             )
-            output_audio = gr.Audio(label="Output Audio", autoplay=True)
+            output_audio = gr.Audio(label="Output Audio", autoplay=True, streaming=True)
         chatbot = gr.Chatbot(label="Conversation", type="messages")
 
     state = gr.State(AppState())