Update app.py

app.py

@@ -7,6 +7,7 @@ import openai
 import gradio as gr
 from dotenv import load_dotenv
 import pinecone
+from chat_utils import get_response, reset_chat_history
 
 """
 login to gcp
@@ -68,10 +69,13 @@ def text2speech(text):
 define voice -> gpt -> text -> voice workflow
 """
 def transcribe(audio):
-    messages = chat.generate_dialogue(audio)
-    voice_path = chat.synthesize_voice(messages)
+    reset_chat_history()
+    voice_path = get_response(audio)
+    messages = get_response(audio, return_messages=True)
     chat_text = "\n".join([f"{msg['role']}: {msg['content']}" for msg in messages])
-    return voice_path, chat_text
+    with open(voice_path, 'rb') as f:
+        voice_bytes = f.read()
+    return voice_bytes, chat_text
     #global messages
 
     """
@@ -128,12 +132,12 @@ def transcribe(audio):
 
 output_text = gr.outputs.Textbox(label="Chat Messages")
 
-output_audio = gr.outputs.Audio(label="Synthesized Voice")
-output_text = gr.outputs.Textbox(label="Chat Messages")
-gr.Interface(
-    fn=transcribe, 
-    inputs=gr.inputs.Audio(source="microphone", label="Speak here..."), 
-    outputs=[output_audio, output_text], 
-    live=True,
-    allow_flagging='never'
-).launch()
+audio_input = gr.inputs.Audio(source="microphone", type="filepath", label="Speak here...")
+chat_output = gr.outputs.Textbox(label="Chat Messages")
+audio_output = gr.outputs.Audio(type="bytes", label="Synthesized Voice")
+
+gr.Interface(fn=transcribe,
+             inputs=audio_input,
+             outputs=[audio_output, chat_output],
+             live=True,
+             allow_flagging=False).launch()