Thanks for the amazing package, you all. I believe the documentation here requires a tensor of shape (nb, nt, nch), not (nb, nch, nt). https://github.com/descriptinc/audiotools/blob/348ebf2034ce24e2a91a553e3171cb00c0c71678/audiotools/core/loudness.py#L182