Update app.py

app.py

@@ -340,6 +340,35 @@ chain_neo4j = (
 
 
 
+# def bot(history, choice, tts_choice, retrieval_mode, model_choice):
+#     if not history:
+#         return history
+    
+#     # Select the model
+#     selected_model = chat_model if model_choice == "GPT-4o" else phi_pipe
+    
+#     response, addresses = generate_answer(history[-1][0], choice, retrieval_mode, selected_model)
+#     history[-1][1] = ""
+
+#     with concurrent.futures.ThreadPoolExecutor() as executor:
+#         if tts_choice == "Alpha":
+#             audio_future = executor.submit(generate_audio_elevenlabs, response)
+#         elif tts_choice == "Beta":
+#             audio_future = executor.submit(generate_audio_parler_tts, response)
+#         # elif tts_choice == "Gamma":
+#         #     audio_future = executor.submit(generate_audio_mars5, response)
+
+#         for character in response:
+#             history[-1][1] += character
+#             time.sleep(0.05)
+#             yield history, None
+
+#         audio_path = audio_future.result()
+#         yield history, audio_path
+
+#     history.append([response, None])
+
+
 def bot(history, choice, tts_choice, retrieval_mode, model_choice):
     if not history:
         return history
@@ -352,24 +381,20 @@ def bot(history, choice, tts_choice, retrieval_mode, model_choice):
 
     with concurrent.futures.ThreadPoolExecutor() as executor:
         if tts_choice == "Alpha":
-            audio_future = executor.submit(generate_audio_elevenlabs, response)
+            audio_stream = generate_audio_elevenlabs(response)
         elif tts_choice == "Beta":
-            audio_future = executor.submit(generate_audio_parler_tts, response)
-        # elif tts_choice == "Gamma":
-        #     audio_future = executor.submit(generate_audio_mars5, response)
+            audio_stream = generate_audio_parler_tts(response)
 
-        for character in response:
-            history[-1][1] += character
-            time.sleep(0.05)
-            yield history, None
-
-        audio_path = audio_future.result()
-        yield history, audio_path
+        # Stream audio chunks in real-time
+        for audio_chunk_path in audio_stream:
+            history[-1][1] += response[:len(history[-1][1]) + 1]  # Simulate text being generated
+            yield history, audio_chunk_path
 
     history.append([response, None])
 
 
 
+
 phi_custom_template = """
 <|system|>
 You are a helpful assistant who provides clear, organized, crisp and conversational responses about an events,concerts,sports and all other activities of Birmingham,Alabama .<|end|>
@@ -934,26 +959,23 @@ def generate_audio_parler_tts(text):
         for new_audio in streamer:
             if new_audio.shape[0] == 0:
                 break
-            # Save or process each audio chunk as it is generated
-            yield sampling_rate, new_audio
+            # Save each audio chunk as it is generated
+            temp_audio_path = os.path.join(tempfile.gettempdir(), f"parler_tts_audio_chunk_{len(audio_segments)}.wav")
+            write_wav(temp_audio_path, sampling_rate, new_audio.astype(np.float32))
+            yield temp_audio_path
 
     audio_segments = []
-    for (sampling_rate, audio_chunk) in generate(text, description, chunk_size_in_s):
-        audio_segments.append(audio_chunk)
-            
-        temp_audio_path = os.path.join(tempfile.gettempdir(), f"parler_tts_audio_chunk_{len(audio_segments)}.wav")
-        write_wav(temp_audio_path, sampling_rate, audio_chunk.astype(np.float32))
-        logging.debug(f"Saved chunk to {temp_audio_path}")
-        
+    for audio_chunk_path in generate(text, description, chunk_size_in_s):
+        audio_segments.append(audio_chunk_path)
+        yield audio_chunk_path  # Yield each audio chunk path immediately
 
     # Combine all the audio chunks into one audio file
     combined_audio = np.concatenate(audio_segments)
     combined_audio_path = os.path.join(tempfile.gettempdir(), "parler_tts_combined_audio_stream.wav")
-
     write_wav(combined_audio_path, sampling_rate, combined_audio.astype(np.float32))
 
     logging.debug(f"Combined audio saved to {combined_audio_path}")
-    return combined_audio_path
+