~VjiT'ddlmZmZddlZdgZdejdejfdZGddejjZ Gd d ejjZ Gd d ejjZ Gd dejjZ dS))OptionalTupleN Conformerlengthsreturnc|jd}ttj|}tj||j|j||| dk}|S)Nr)devicedtype) shapeinttorchmaxitemaranger r expand unsqueeze)r batch_size max_length padding_masks U/root/voice-cloning/.venv/lib/python3.11/site-packages/torchaudio/models/conformer.py_lengths_to_padding_maskr s~q!JUYw'',,..//J< 7>WWW^^J   1  L cleZdZdZ ddededededed ed d ffd Zd ej d ej fdZ xZ S)_ConvolutionModuleaConformer convolution module. Args: input_dim (int): input dimension. num_channels (int): number of depthwise convolution layer input channels. depthwise_kernel_size (int): kernel size of depthwise convolution layer. dropout (float, optional): dropout probability. (Default: 0.0) bias (bool, optional): indicates whether to add bias term to each convolution layer. (Default: ``False``) use_group_norm (bool, optional): use GroupNorm rather than BatchNorm. (Default: ``False``) F input_dim num_channelsdepthwise_kernel_sizedropoutbiasuse_group_normrNc.t|dz dzdkrtdtj||_tjtj|d|zddd|tj dtj|||d|dz dz|||r!tj d|ntj |tj tj||ddd| tj ||_dS) Nr rz