o i@srddlmZddlmZmZmZddlZddlmZddl mm Z ddl m Z ddlmZGdddeZdS)) OrderedDict)ListTupleUnionN) ComplexTensor)AbsMaskcseZdZ   ddededeffdd Zed efd d Zd eej e fd ej dej ded e e eej e fej e ff ddZZS) MultiMaskrelu input_dimbottleneck_dim max_num_spkc sVt||_||_t|_td|dD]}|jtj |||dddqdS)aMultiple 1x1 convolution layer Module. This module corresponds to the final 1x1 conv block and non-linear function in TCNSeparator. This module has multiple 1x1 conv blocks. One of them is selected according to the given num_spk to handle flexible num_spk. Args: input_dim: Number of filters in autoencoder bottleneck_dim: Number of channels in bottleneck 1 * 1-conv block max_num_spk: Number of mask_conv1x1 modules (>= Max number of speakers in the dataset) mask_nonlinear: use which non-linear function to generate mask F)biasN) super__init__ _max_num_spkmask_nonlinearnn ModuleList mask_conv1x1rangeappendConv1d)selfr r rrz __class__R/home/ubuntu/.local/lib/python3.10/site-packages/espnet2/diar/layers/multi_mask.pyrs  zMultiMask.__init__returncCs|jS)N)r)rrrr r0szMultiMask.max_num_spkinputilensbottleneck_featnum_spkcsD\}}}|dd}|j|d|}t|jD]"} | |dkr=|dtj|j| |dd||ddd7}q|||||}|jdkrStj |dd} n%|jdkr^t |} n|jdkrit |} n|jd krtt |} ntd | dd } | jdd} fd d | D} ttdd tt| D| } | || fS)aKeep this API same with TasNet. Args: input: [M, K, N], M is batch size ilens (torch.Tensor): (M,) bottleneck_feat: [M, K, B] num_spk: number of speakers (Training: oracle, Inference: estimated by other module (e.g, EEND-EDA)) Returns: masked (List[Union(torch.Tensor, ComplexTensor)]): [(M, K, N), ...] ilens (torch.Tensor): (M,) others predicted data, e.g. masks: OrderedDict[ 'mask_spk1': torch.Tensor(Batch, Frames, Freq), 'mask_spk2': torch.Tensor(Batch, Frames, Freq), ... 'mask_spkn': torch.Tensor(Batch, Frames, Freq), ] rg)sizesoftmax)dimr sigmoidtanhz$Unsupported mask non-linear functionr csg|]}|qSrr).0mr"rr lsz%MultiMask.forward..cSsg|] }d|dqS)z mask_spk{}r)format)r,irrr r/os)r' transposerrrF interpolateviewrr(r torchr*r+ ValueErrorunbindrziplen)rr"r#r$r%MKNscorerest_maskmasksmaskedothersrr.r forward4s>             zMultiMask.forward)r r r )__name__ __module__ __qualname__intrpropertyrrr6TensorrrrrrC __classcell__rrrr rs0  r) collectionsrtypingrrrr6torch.nnrtorch.nn.functional functionalr3torch_complex.tensorrespnet2.diar.layers.abs_maskrrrrrr s