Update app.py

app.py

@@ -4,17 +4,24 @@ import gradio as gr
 # Load pre-trained models
 stt_model = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
 nlp_model = pipeline("text-generation", model="sshleifer/tiny-gpt2")
-tts_model = pipeline("text-to-speech", model="facebook/fastspeech2-en-ljspeech")
-
+tts_model = pipeline("text-to-speech", model="coqui/XTTS-v2")
 
 # Define a function to handle the workflow
 def conversation(audio):
     # Step 1: Convert speech to text
     text = stt_model(audio)["text"]
-    # Step 2: Generate a response
-    response = nlp_model(text, max_length=50)[0]["generated_text"]
-    # Step 3: Convert response text to speech
+    
+    # Step 2: Generate a response (contextual supermarket-related training)
+    if "supermarket" in text.lower():
+        # Simple supermarket-based response; this can be expanded with more specific data
+        response = "Are you looking for something in particular at the supermarket?"
+    else:
+        # Default response generation (using GPT-2 model)
+        response = nlp_model(text, max_length=50)[0]["generated_text"]
+    
+    # Step 3: Convert response text to speech using XTTS-v2
     audio_response = tts_model(response)
+    
     return text, response, audio_response
 
 # Create Gradio Interface