Update app.py

app.py

@@ -4,15 +4,14 @@ import google.generativeai as genai
 import speech_recognition as sr
 from tempfile import NamedTemporaryFile
 from dotenv import load_dotenv
-import torch
-from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM
+import torchaudio  # <-- Needed to save audio
+from transformers import pipeline
 
-# Load environment variables
+# Load environment variables (if using .env file)
 load_dotenv()
-#API_KEY = "AIzaSyB3N9BHeIWs_8sdFK76PU-v9N6prcIq2Hw" #os.getenv("API_KEY")  # or hardcode as "your_gemini_api_key"
 
 # Configure Gemini
-genai.configure(api_key="AIzaSyB3N9BHeIWs_8sdFK76PU-v9N6prcIq2Hw") #=API_KEY)
+genai.configure(api_key="AIzaSyB3N9BHeIWs_8sdFK76PU-v9N6prcIq2Hw")
 gemini = genai.GenerativeModel("gemini-1.5-pro")
 
 # Load YarnGPT as text-to-speech pipeline
@@ -28,15 +27,13 @@ def transcribe_audio(audio_path):
         audio_data = recognizer.record(source)
         try:
             return recognizer.recognize_google(audio_data)
-        except sr.UnknownValueError:
-            return ""
-        except sr.RequestError:
+        except (sr.UnknownValueError, sr.RequestError):
             return ""
 
 # Main AI interaction
 def chat_with_ai(audio, text_input, emotion):
     user_text = text_input or ""
-    
+
     if audio:
         transcribed = transcribe_audio(audio)
         if transcribed:
@@ -47,14 +44,16 @@ def chat_with_ai(audio, text_input, emotion):
     if not user_text.strip():
         return "No input provided.", None
 
-    # Prepare emotion-aware prompt
-    prompt = f"The user is feeling {emotion}. Respond appropriately to help them feel better.\nUser said: {user_text}"
+    # Emotion-aware prompt
+    prompt = f"The user is feeling {emotion}. Respond supportively and help them feel better.\nUser said: {user_text}"
     ai_response = gemini.generate_content(prompt).text
 
-    # Generate audio using YarnGPT
-    tts_output = tts_pipeline(ai_response)
-    audio_path = NamedTemporaryFile(delete=False, suffix=".wav").name
-    torchaudio.save(audio_path, tts_output["audio"], sample_rate=tts_output["sampling_rate"])
+    try:
+        tts_output = tts_pipeline(ai_response)
+        audio_path = NamedTemporaryFile(delete=False, suffix=".wav").name
+        torchaudio.save(audio_path, tts_output["audio"], sample_rate=tts_output["sampling_rate"])
+    except Exception as e:
+        return ai_response, None
 
     return ai_response, audio_path
 
@@ -64,7 +63,7 @@ with gr.Blocks(title="Mind AID AI Assistant") as iface:
 
     with gr.Row():
         emotion = gr.Dropdown(label="Select Your Emotional State", choices=emotion_options, value="neutral")
-    
+
     with gr.Row():
         text_input = gr.Textbox(label="Or type your message here (optional)", lines=2)
         audio_input = gr.Audio(type="filepath", label="Or speak to the AI")