modify app

app.py

@@ -103,16 +103,19 @@ def perform_ito(input_audio, reference_audio, ito_reference_audio, num_steps, op
             current_output = current_output.cpu().numpy()
         
         # Normalize output audio
-        current_output = loudness_normalize(current_output, args.sample_rate)
+        # current_output = loudness_normalize(current_output, args.sample_rate)
         
         # Denormalize the audio to int16
         current_output = denormalize_audio(current_output, dtype=np.int16)
-        
-        # Ensure the audio is in the correct shape (samples, channels)
+            
         if current_output.ndim == 1:
             current_output = current_output.reshape(-1, 1)
         elif current_output.ndim > 2:
             current_output = current_output.squeeze()
+
+        # Ensure the audio is in the correct shape (samples, channels)
+        if current_output.shape[1] > current_output.shape[0]:
+            current_output = current_output.transpose(1,0)
         
         yield (args.sample_rate, current_output), ito_param_output, step, ito_log, loss_values