Update app.py

app.py

@@ -194,8 +194,11 @@ def generate_audio_from_image(image):
     # Convert the generated spectrogram to audio
     generated_audio = spectrogram_to_audio(generated_spectrogram.squeeze(0).cpu())
 
-    # Return the audio as (sample_rate, audio) tuple for Gradio to process
-    return (sample_rate, generated_audio)
+    # Convert the audio tensor to a NumPy array before passing it to Gradio
+    generated_audio_numpy = generated_audio.numpy()
+
+    # Return the sample rate and the NumPy array containing the audio
+    return (sample_rate, generated_audio_numpy)
 
 
 # Gradio Interface