Update app.py

app.py

@@ -16,10 +16,13 @@ print("Initializing Whisper model...")
 model = whisper.load_model("base") 
 print("System Ready.")
 
+# The core instructions for the AI
+SYSTEM_PROMPT = {"role": "system", "content": "You are a professional, intelligent AI assistant demonstrating a low-latency voice architecture. Provide concise, highly accurate, and polite responses."}
+
 # 2. Main Processing Logic
-def process_voice_conversation(audio_path, history):
+def process_voice_conversation(audio_path, chat_history, llm_state):
     if not audio_path:
-        return history, None, None
+        return chat_history, llm_state, None, None
 
     try:
         # Step A: Speech-to-Text
@@ -27,37 +30,43 @@ def process_voice_conversation(audio_path, history):
         user_text = transcription["text"].strip()
         
         if not user_text:
-            return history, None, None
+            return chat_history, llm_state, None, None
 
-        history.append({"role": "user", "content": user_text})
+        # Add to AI's internal memory
+        llm_state.append({"role": "user", "content": user_text})
 
         # Step B: LLM Processing via Groq
-        messages = [
-            {"role": "system", "content": "You are a professional, intelligent AI assistant demonstrating a low-latency voice architecture. Provide concise, highly accurate, and polite responses."}
-        ]
-        messages.extend(history)
-
         chat_completion = client.chat.completions.create(
-            messages=messages,
+            messages=llm_state,
             model="llama-3.3-70b-versatile",
         )
         ai_text = chat_completion.choices[0].message.content
-        history.append({"role": "assistant", "content": ai_text})
+        
+        # Add response to AI's internal memory
+        llm_state.append({"role": "assistant", "content": ai_text})
+
+        # Add the conversation pair to the UI Chatbot
+        chat_history.append((user_text, ai_text))
 
         # Step C: Text-to-Speech
         tts = gTTS(text=ai_text, lang='en', slow=False)
         temp_audio = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3")
         tts.save(temp_audio.name)
 
-        return history, temp_audio.name, None
+        # Return UI history, Memory state, Output Audio, and clear Input Audio
+        return chat_history, llm_state, temp_audio.name, None
 
     except Exception as e:
         error_msg = f"System Error: {str(e)}"
-        history.append({"role": "assistant", "content": error_msg})
-        return history, None, None
+        # Display the error in the chat interface safely
+        chat_history.append(("Audio processed...", error_msg))
+        return chat_history, llm_state, None, None
+
+# Function to completely wipe the session memory and UI
+def reset_conversation():
+    return [], [SYSTEM_PROMPT], None, None
 
 # 3. Professional UI Design
-# Using a clean, monochrome theme favored in academic/enterprise tools
 custom_theme = gr.themes.Monochrome(
     font=[gr.themes.GoogleFont("Inter"), "ui-sans-serif", "system-ui", "sans-serif"],
     primary_hue="slate",
@@ -66,6 +75,9 @@ custom_theme = gr.themes.Monochrome(
 
 with gr.Blocks(title="VocaFree AI - Research Prototype", theme=custom_theme) as demo:
     
+    # Hidden state variable to hold the LLM's memory securely
+    llm_state = gr.State([SYSTEM_PROMPT])
+    
     # Header
     gr.Markdown(
         """
@@ -75,16 +87,15 @@ with gr.Blocks(title="VocaFree AI - Research Prototype", theme=custom_theme) as
         """
     )
     
-    # Use Tabs to separate the Demo from the Documentation
     with gr.Tabs():
         
         # TAB 1: The Live App
         with gr.Tab("🎙️ Live Interaction"):
             with gr.Row():
                 with gr.Column(scale=2):
+                    # Removed the problematic 'type="messages"' argument 
                     chatbot = gr.Chatbot(
                         label="Conversation Transcript", 
-                        type="messages", 
                         height=450,
                         avatar_images=(None, "⚙️") # Professional gear icon for the AI
                     )
@@ -105,7 +116,7 @@ with gr.Blocks(title="VocaFree AI - Research Prototype", theme=custom_theme) as
                         autoplay=True, 
                         interactive=False
                     )
-                    clear_btn = gr.ClearButton([chatbot, audio_input, audio_output], value="Reset Session")
+                    clear_btn = gr.Button("🗑️ Reset Session")
 
         # TAB 2: Architecture & Documentation
         with gr.Tab("📊 System Architecture"):
@@ -134,11 +145,18 @@ with gr.Blocks(title="VocaFree AI - Research Prototype", theme=custom_theme) as
         """
     )
 
-    # Event Wiring
+    # Event Wiring: Submit Audio
     submit_btn.click(
         fn=process_voice_conversation,
-        inputs=[audio_input, chatbot],
-        outputs=[chatbot, audio_output, audio_input]
+        inputs=[audio_input, chatbot, llm_state],
+        outputs=[chatbot, llm_state, audio_output, audio_input]
+    )
+    
+    # Event Wiring: Clear Session (Wipes UI and AI Memory)
+    clear_btn.click(
+        fn=reset_conversation,
+        inputs=[],
+        outputs=[chatbot, llm_state, audio_input, audio_output]
     )
 
 if __name__ == "__main__":