9i  SSKrSSKrSSKJr SSKJs Jr "SS\R5r "SS\ 5r S\ S\ S\ S \ 4S jr "S S \R5r "S S\ 5r"SS\RR5rg)Nc`^\rSrSrS U4SjjrSrS\S\S\4SjrSrS r U=r $) Conv1d_O c >>[T U]5 X lXPlX`lXplXlSUlXlUcUc [S5eUcURU5n[R"UUURURURSUU S9Ul g)NFz.Must provide one of input_shape or in_channelsr)stridedilationpaddinggroupsbias)super__init__ kernel_sizerrr padding_mode unsqueezeskip_transpose ValueError_check_input_shapennConv1dconv) self out_channelsr input_shape in_channelsrrr r r rr __class__s _/var/www/html/land-doc-ocr/venv/lib/python3.13/site-packages/modelscope/models/audio/sv/TDNN.pyr Conv1d_O.__init__ s &   (,  ;#6MN N  11+>KII     ;;]]   cUR(dURSS5nUR(aURS5nURS:Xa2UR XR UR UR5nOnURS:Xa5UR S- UR -n[R"XS45nO)URS:XaO[SUR-5eURU5nUR(aURS5nUR(dURSS5nU$)zReturns the output of the convolution. Arguments --------- x : torch.Tensor (batch, time, channel) input to convolve. 2d or 4d tensors are expected. samecausalrvalidz1Padding must be 'same', 'valid' or 'causal'. Got ) r transposerr _manage_paddingrrrFpadrrsqueeze)rxnum_padwxs rforwardConv1d_O.forward3s "" Ar"A >> AA <<6 !$$Q(8(8$--%)[[2A\\X %''!+t}}>AB""a$B rrrrczURSn[XTX#5n[R"XURS9nU$)Nr!)mode)shapeget_padding_elemr'r(r)rr*rrrL_inr s rr&Conv1d_O._manage_padding\s<wwr{#4G EE!4#4#4 5rc&[U5S:Xa SUlSnOLUR(aUSnO5[U5S:XaUSnO [S[ [U55-5eUR S-S:Xa[SUR -5eU$)zIChecks the input shape and returns the number of input channels. Tr z"conv1d expects 2d, 3d inputs. Got rz4The field kernel size must be an odd number. Got %s.)lenrrrstrr)rr1rs rrConv1d_O._check_input_shapens u:?!DNK  (K Z1_(KA"3u:/0 0   a 1 $F!!#$ $r)rrrr rrrr) NNr r r"r TreflectF) __name__ __module__ __qualname____firstlineno__r r-intr&r__static_attributes__ __classcell__rs@rrr sY  & P'R   $rrc(^\rSrSrU4SjrSrU=r$)rc*>[TU]"USS0UD6 gNrTr r rargskwargsrs rr Conv1d.__init__ t>>v>rr<r=r>r?r rArBrCs@rrr ??rrr3rrrcUS:a8[RXU-- U- S-5nXS- -X#--nUS-US-/nU$XUS- -- S- U-S-nX- S-X- S-/nU$)zThis function computes the number of elements to add for zero-padding. Arguments --------- L_in : int stride: int kernel_size : int dilation : int r r6)mathceil)r3rrrn_stepsL_outr s rr2r2sz))d8%;;vEJKA+&)??!#[A%56 NK!O44q8VCaGLQ&!(;< NrcB^\rSrSrSU4SjjrSrSrU=r$) BatchNorm1d_Oc >[T U]5 XplXlUc U(aUSnOUcUSn[R "UUUUUS9Ulg)Nr r!)epsmomentumaffinetrack_running_stats)r r combine_batch_timerr BatchNorm1dnorm) rr input_sizerZr[r\r]r^rrs rr BatchNorm1d_O.__init__s[ "4,  .$QJ  $RJNN  3   rcURnUR(aRURS:XaURUSUS-US5nOFURUSUS-USUS5nO#UR(dUR SS5nUR U5nUR(aURU5nU$UR(dUR SS5nU$)zReturns the normalized input tensor. Arguments --------- x : torch.Tensor (batch, time, [channels]) input to normalize. 2d or 3d tensors are expected in input 4d tensors can be used when combine_dims=True. r7rr r6r!)r1r^ndimreshaperr%r`)rr*shape_orx_ns rr-BatchNorm1d_O.forwards77  " "vv{IIhqkHQK7!EIIhqkHQK7!&qk+$$ B"Aiil  " "++h'C $$--2&C r)r^r`r)NNgh㈵>g?TTFF)r<r=r>r?r r-rArBrCs@rrWrWs,    8rrWc(^\rSrSrU4SjrSrU=r$)r_c*>[TU]"USS0UD6 grGrHrIs rr BatchNorm1d.__init__rMrrNrOrCs@rr_r_rPrr_c~^\rSrSrSrS\R RS/SQ/SQ/SQSS 4U4S jjrS S jr S r U=r $)XvectoraThis model extracts X-vectors for speaker recognition and diarization. Arguments --------- device : str Device used e.g. "cpu" or "cuda". activation : torch class A class for constructing the activation layers. tdnn_blocks : int Number of time-delay neural (TDNN) layers. tdnn_channels : list of ints Output channels for TDNN layer. tdnn_kernel_sizes : list of ints List of kernel sizes for each TDNN layer. tdnn_dilations : list of ints List of dilations for kernels in each TDNN layer. lin_neurons : int Number of neurons in linear layers. Example ------- >>> compute_xvect = Xvector('cpu') >>> input_feats = torch.rand([5, 10, 40]) >>> outputs = compute_xvect(input_feats) >>> outputs.shape torch.Size([5, 1, 512]) cpu)rrrrrri)rqr7r7r r )r r6r7r r rrPc >[T U]5 [R"5Ul[ U5HCn XIn URR [UU XYXiS9U"5[U S9/5 XInME g)N)rrrr)ra) r r r ModuleListblocksrangeextendrr_) rdevice activation tdnn_blocks tdnn_channelstdnn_kernel_sizestdnn_dilations lin_neuronsr block_indexrrs rr Xvector.__init__s mmo !-K(5L KK   +!- 1 >+8   |4  (4K.rcURSS5nURH nU"XS9nM URSS5nU$![a U"U5nM7f=f)z=Returns the x-vectors. Arguments --------- x : torch.Tensor r r6)lengths)r%rv TypeError)rr*lenslayers rr-Xvector.forwardsb KK1 [[E !*! KK1  !H sAAA)rv)N) r<r=r>r?__doc__torchr LeakyReLUr r-rArBrCs@rrnrns;<88%%0)&5<rrn)numpynprtorch.nnrtorch.nn.functional functionalr'Modulerrr@r2rWr_rnrNrrrs yryyyz?X? 3#,9BII9x?-? KehhooKr