Update app.py

app.py

@@ -447,25 +447,23 @@ import concurrent.futures
 import asyncio
 
 async def bot(history, choice, tts_choice, retrieval_mode, model_choice):
-    # Initialize an empty response
     response = ""
 
-    # Start generating the text and audio in parallel
-    text_future = asyncio.create_task(generate_text(history, choice, retrieval_mode, model_choice))
+    # Start generating text asynchronously
+    text_gen = generate_text(history, choice, retrieval_mode, model_choice)
     audio_future = None
 
-    while not text_future.done():
-        # Stream the text as it's being generated
-        chunk = await text_future
+    # Iterate over the text generator
+    async for chunk in text_gen:
         response += chunk
         history[-1][1] += chunk
         yield history, None  # Stream the text output as it's generated
 
-        # Start generating Parler TTS if selected
+        # Start generating Parler TTS if selected and not started already
         if tts_choice == "Beta" and audio_future is None:
             audio_future = asyncio.create_task(generate_audio_parler_tts(response, callback=lambda audio_chunk: yield_audio(audio_chunk)))
 
-    # Wait for the audio to finish streaming
+    # Wait for the audio to finish streaming if it was started
     if audio_future is not None:
         await audio_future
 
@@ -475,11 +473,13 @@ def yield_audio(audio_chunk):
     write_wav(temp_audio_path, 16000, audio_chunk.astype(np.float32))
     return temp_audio_path
 
+# Text generator as an async generator
 async def generate_text(history, choice, retrieval_mode, model_choice):
     # Simulate text generation chunk by chunk
-    for char in "Generating text response...":
+    text_to_generate = "Generating text response..."
+    for char in text_to_generate:
         await asyncio.sleep(0.05)  # Simulate time delay between character generation
-        yield char
+        yield char  # Yield each character as it's generated