Update app.py

app.py

@@ -299,21 +299,24 @@ def voice_conversion(input, reference, steps, guidance, pitch, speed):
         vc_target = inference_module.cfm.inference(cat_condition, torch.LongTensor([cat_condition.size(1)]).to(mel2.device), mel2, style2, None, steps, inference_cfg_rate=guidance)
         vc_target = vc_target[:, :, mel2.size(2):]
         print(f"[INFO] | vc_target shape: {vc_target.shape}")
-        print(vc_target)
+
+        # TEMP
+        output_wave = vc_target[0].cpu().numpy()
+        generated_wave_chunks.append(output_wave)
+        
         # Generate waveform using BigVGAN
+        """
         vc_wave = bigvgan_fn(vc_target.float())[0]
         print(f"[INFO] | vc_wave shape: {vc_wave.shape}")
-        print(vc_wave)
+        
         # Handle the generated waveform
         output_wave = vc_wave[0].cpu().numpy()
         generated_wave_chunks.append(output_wave)
+        """
         
         # Ensure processed_frames increments correctly to avoid infinite loop
-        print(processed_frames)
-        print(cond.size(1))
         processed_frames += vc_target.size(2)
-        print(processed_frames)
-        print(cond.size(1))
+        
         print(f"[INFO] | Processed frames updated to: {processed_frames}")
     
     # Concatenate all generated wave chunks