Model compression

wonder to make my model size smaller, and speed fast, which is sepformer with WSJ0-3mix datasets. is there any method or code to compress the model in SpeechBrain? thanks! Or any discussion will be welcomed.