Update app.py

app.py

@@ -10,6 +10,7 @@ import torch
 import random
 from openai import OpenAI
 import subprocess
+from starlette.requests import ClientDisconnect
 
 LLAMA_3B_API_ENDPOINT = os.environ.get("LLAMA_3B_API_ENDPOINT")
 LLAMA_3B_API_KEY = os.environ.get("LLAMA_3B_API_KEY")
@@ -146,30 +147,37 @@ def translate_speech(audio_file, target_language):
         return None
 
 async def respond(audio, model, seed, target_language):
-    if audio is None:
-        return None, None
-    
-    user_input = transcribe(audio)
-    if not user_input:
-        return None, None
-    
-    if user_input.lower().startswith("please translate"):
-        # Extract the actual content to translate
-        content_to_translate = user_input[len("please translate"):].strip()
-        translated_audio = translate_speech(audio, target_language)
-        return None, translated_audio
-    else:
-        reply = models(user_input, model, seed)
-        communicate = edge_tts.Communicate(reply, voice="en-US-ChristopherNeural")
-        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
-            tmp_path = tmp_file.name
-            await communicate.save(tmp_path)
-        return tmp_path, None
+    try:
+        if audio is None:
+            return None, None, "No input detected."
+        
+        user_input = transcribe(audio)
+        if not user_input:
+            return None, None, "Could not transcribe audio."
+        
+        if user_input.lower().startswith("please translate"):
+            # Extract the actual content to translate
+            content_to_translate = user_input[len("please translate"):].strip()
+            translated_audio = translate_speech(audio, target_language)
+            return None, translated_audio, f"Translated to {target_language}"
+        else:
+            reply = models(user_input, model, seed)
+            communicate = edge_tts.Communicate(reply, voice="en-US-ChristopherNeural")
+            with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
+                tmp_path = tmp_file.name
+                await communicate.save(tmp_path)
+            return tmp_path, None, "Voice assistant response"
+    except ClientDisconnect:
+        print("Client disconnected")
+        return None, None, "Client disconnected. Please try again."
+    except Exception as e:
+        print(f"An error occurred: {str(e)}")
+        return None, None, f"An error occurred: {str(e)}"
 
 def clear_history():
     global conversation_history
     conversation_history = []
-    return None, None
+    return None, None, "Conversation history cleared."
 
 with gr.Blocks(css="style.css") as demo:
     description = gr.Markdown("# <center><b>Optimus Prime: Voice Assistant with Translation</b></center>")
@@ -200,17 +208,17 @@ with gr.Blocks(css="style.css") as demo:
     input_audio = gr.Audio(label="User Input", sources=["microphone"], type="filepath")
     output_audio = gr.Audio(label="AI Response", type="filepath", interactive=False, autoplay=True)
     translated_audio = gr.Audio(label="Translated Audio", type="filepath", interactive=False, autoplay=True)
+    status_message = gr.Textbox(label="Status", interactive=False)
 
     clear_button = gr.Button("Clear Conversation History")
 
-    gr.Interface(
+    input_audio.change(
         fn=respond, 
         inputs=[input_audio, select, seed, target_lang],
-        outputs=[output_audio, translated_audio],
-        live=True
+        outputs=[output_audio, translated_audio, status_message],
     )
 
-    clear_button.click(fn=clear_history, inputs=[], outputs=[output_audio, translated_audio])
+    # clear_button.click(fn=clear_history, inputs=[], outputs=[output_audio, translated_audio, status_message])
 
 if __name__ == "__main__":
     demo.queue(max_size=200).launch()