Update app.py

app.py

@@ -185,6 +185,10 @@ def generate_audio(script_output, voice1, voice2, num_hosts, temperature, top_p,
             with torch.no_grad():
                 paragraph_audio = snac_model(codes_tensor)
             
+            # Handle tuple output
+            if isinstance(paragraph_audio, tuple):
+                paragraph_audio = paragraph_audio[0]  # Assume the first element is the audio tensor
+            
             paragraph_audio = paragraph_audio.cpu().numpy().flatten()
             
             silences = detect_silence(paragraph_audio)