o ei)@sdZddlZddlmZddlZddlmZddlm Z ddl m Z ddl m Z GdddejjZGd d d ejjjZGd d d ejjjZdS) zqA popular speaker recognition and diarization model. Authors * Nauman Dawalatabad 2020 * Mirco Ravanelli 2020 N)Conv1d)Linear) BatchNorm1d)StatisticsPoolingcsLeZdZdZdejjdgdgdgdddffd d Zdd d ZZ S)XvectoraThis model extracts X-vectors for speaker recognition and diarization. Arguments --------- device : str Device used e.g. "cpu" or "cuda". activation : torch class A class for constructing the activation layers. tdnn_blocks : int Number of time-delay neural (TDNN) layers. tdnn_channels : list of ints Output channels for TDNN layer. tdnn_kernel_sizes : list of ints List of kernel sizes for each TDNN layer. tdnn_dilations : list of ints List of dilations for kernels in each TDNN layer. lin_neurons : int Number of neurons in linear layers. in_channels : int Expected size of input features. Example ------- >>> compute_xvect = Xvector('cpu') >>> input_feats = torch.rand([5, 10, 40]) >>> outputs = compute_xvect(input_feats) >>> outputs.shape torch.Size([5, 1, 512]) cpu)r r r i)rr r )r r r r r (c stt|_t|D]!} || } |jt|| || || d|t| dg|| }q|j t |j t | d|ddddS)N) in_channels out_channels kernel_sizedilation) input_sizer TF)r n_neuronsbias combine_dims) super__init__nn ModuleListblocksrangeextendrrappendrr) selfdevice activation tdnn_blocks tdnn_channelstdnn_kernel_sizestdnn_dilations lin_neuronsr block_indexr __class__^/home/ubuntu/transcripts/venv/lib/python3.10/site-packages/speechbrain/lobes/models/Xvector.pyr2s2   zXvector.__init__Nc Cs:|jD]}z|||d}Wqty||}Yqw|S)a>Returns the x-vectors. Arguments --------- x : torch.Tensor Inputs features for extracting x-vectors. lens : torch.Tensor The corresponding relative lengths of the inputs. Returns ------- x : torch.Tensor X-vectors. )lengths)r TypeError)rxlenslayerr)r)r*forward^s   zXvector.forward)N) __name__ __module__ __qualname____doc__torchr LeakyReLUrr0 __classcell__r)r)r'r*rs ,rcs.eZdZdZejjdddffdd ZZS) ClassifieraThis class implements the last MLP on the top of xvector features. Arguments --------- input_shape : tuple Expected shape of an example input. activation : torch class A class for constructing the activation layers. lin_blocks : int Number of linear layers. lin_neurons : int Number of neurons in linear layers. out_neurons : int Number of output neurons. Example ------- >>> input_feats = torch.rand([5, 10, 40]) >>> compute_xvect = Xvector() >>> xvects = compute_xvect(input_feats) >>> classify = Classifier(input_shape=xvects.shape) >>> output = classify(xvects) >>> output.shape torch.Size([5, 1, 1211]) r r icstj|d|j|dd|jtjjjdd|dkr'|jtjjjddt |D]9}d|}|j jtjjj|d|j |jtjj j |dd d |j |j|dd|j |jtjjjddq+|jtjj j |d d |jtjj jdd dddS)N input_shapeact layer_namenormrDNNblock_Tlinear)rrr=outrr=) apply_logsoftmax)rrrsbnnet normalizationr containers Sequentialrr?rAr activationsSoftmaxrr:r lin_blocksr% out_neuronsr& block_namer'r)r*rs4        zClassifier.__init__ r1r2r3r4r5rr6rr7r)r)r'r*r8vr8cs.eZdZdZejjdddffdd ZZS) DiscriminatoraThis class implements a discriminator on the top of xvector features. Arguments --------- input_shape : tuple Expected shape of the input tensor. activation : torch class A class for constructing the activation layers. lin_blocks : int Number of linear layers. lin_neurons : int Number of neurons in linear layers. out_neurons : int Size of the output vector. Example ------- >>> input_feats = torch.rand([5, 10, 40]) >>> compute_xvect = Xvector() >>> xvects = compute_xvect(input_feats) >>> discriminate = Discriminator(xvects.shape) >>> output = discriminate(xvects) >>> output.shape torch.Size([5, 1, 1]) r r cstj|d|dkr|jtjjjddt|D]:}d|}|jjtjjj|d|j|jtjj j |dddd |j|jtjj j d d|j|j|d dq|jtjj j |d d dS)Nr9rr?r<r@TFrA)rrrr=r>r;rBrC) rrrrFrGrIrJrr?rArrHrrMr'r)r*rs,        zDiscriminator.__init__rQr)r)r'r*rSrRrS)r4r5torch.nnr speechbrainrFspeechbrain.nnet.CNNrspeechbrain.nnet.linearrspeechbrain.nnet.normalizationrspeechbrain.nnet.poolingrModulerrGrIrJr8rSr)r)r)r*s     cD