update

bigvgan.py

@@ -324,7 +324,7 @@ class BigVGAN(
             print(f"[WARNING] You have specified use_cuda_kernel=True during BigVGAN.from_pretrained(). Only inference is supported (training is not implemented)!")
             print(f"[WARNING] You need nvcc and ninja installed in your system that matches your PyTorch build is using to build the kernel. If not, the model will fail to initialize or generate incorrect waveform!")
             print(f"[WARNING] For detail, see the official GitHub repository: https://github.com/NVIDIA/BigVGAN?tab=readme-ov-file#using-custom-cuda-kernel-for-synthesis")
-            model = cls(h, use_cuda_kernel=use_cuda_kernel)
+        model = cls(h, use_cuda_kernel=use_cuda_kernel)
 
         ##################################################################
         # download and load pretrained generator weight