paddlespeech.t2s.models.wavernn.wavernn module

class paddlespeech.t2s.models.wavernn.wavernn.MelResNet(res_blocks: int = 10, compute_dims: int = 128, res_out_dims: int = 128, aux_channels: int = 80, aux_context_window: int = 0)

Bases: Layer

Methods

__call__(*inputs, **kwargs)	Call self as a function.
add_parameter(name, parameter)	Adds a Parameter instance.
add_sublayer(name, sublayer)	Adds a sub Layer instance.
apply(fn)	Applies fn recursively to every sublayer (as returned by .sublayers()) as well as self.
buffers([include_sublayers])	Returns a list of all buffers from current layer and its sub-layers.
children()	Returns an iterator over immediate children layers.
clear_gradients()	Clear the gradients of all parameters for this layer.
create_parameter(shape[, attr, dtype, ...])	Create parameters for this layer.
create_tensor([name, persistable, dtype])	Create Tensor for this layer.
create_variable([name, persistable, dtype])	Create Tensor for this layer.
eval()	Sets this Layer and all its sublayers to evaluation mode.
extra_repr()	Extra representation of this layer, you can have custom implementation of your own layer.
forward(x)	Args:
full_name()	Full name for this layer, composed by name_scope + "/" + MyLayer.__class__.__name__
load_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
named_buffers([prefix, include_sublayers])	Returns an iterator over all buffers in the Layer, yielding tuple of name and Tensor.
named_children()	Returns an iterator over immediate children layers, yielding both the name of the layer as well as the layer itself.
named_parameters([prefix, include_sublayers])	Returns an iterator over all parameters in the Layer, yielding tuple of name and parameter.
named_sublayers([prefix, include_self, ...])	Returns an iterator over all sublayers in the Layer, yielding tuple of name and sublayer.
parameters([include_sublayers])	Returns a list of all Parameters from current layer and its sub-layers.
register_buffer(name, tensor[, persistable])	Registers a tensor as buffer into the layer.
register_forward_post_hook(hook)	Register a forward post-hook for Layer.
register_forward_pre_hook(hook)	Register a forward pre-hook for Layer.
set_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
set_state_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
state_dict([destination, include_sublayers, ...])	Get all parameters and persistable buffers of current layer and its sub-layers.
sublayers([include_self])	Returns a list of sub layers.
to([device, dtype, blocking])	Cast the parameters and buffers of Layer by the give device, dtype and blocking.
to_static_state_dict([destination, ...])	Get all parameters and buffers of current layer and its sub-layers.
train()	Sets this Layer and all its sublayers to training mode.

backward
register_state_dict_hook

forward(x)

Args:

x (Tensor):

Input tensor (B, in_dims, T).

Returns:

Tensor:

Output tensor (B, res_out_dims, T).

class paddlespeech.t2s.models.wavernn.wavernn.ResBlock(dims)

Bases: Layer

Methods

__call__(*inputs, **kwargs)	Call self as a function.
add_parameter(name, parameter)	Adds a Parameter instance.
add_sublayer(name, sublayer)	Adds a sub Layer instance.
apply(fn)	Applies fn recursively to every sublayer (as returned by .sublayers()) as well as self.
buffers([include_sublayers])	Returns a list of all buffers from current layer and its sub-layers.
children()	Returns an iterator over immediate children layers.
clear_gradients()	Clear the gradients of all parameters for this layer.
create_parameter(shape[, attr, dtype, ...])	Create parameters for this layer.
create_tensor([name, persistable, dtype])	Create Tensor for this layer.
create_variable([name, persistable, dtype])	Create Tensor for this layer.
eval()	Sets this Layer and all its sublayers to evaluation mode.
extra_repr()	Extra representation of this layer, you can have custom implementation of your own layer.
forward(x)	conv -> bn -> relu -> conv -> bn + residual connection
full_name()	Full name for this layer, composed by name_scope + "/" + MyLayer.__class__.__name__
load_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
named_buffers([prefix, include_sublayers])	Returns an iterator over all buffers in the Layer, yielding tuple of name and Tensor.
named_children()	Returns an iterator over immediate children layers, yielding both the name of the layer as well as the layer itself.
named_parameters([prefix, include_sublayers])	Returns an iterator over all parameters in the Layer, yielding tuple of name and parameter.
named_sublayers([prefix, include_self, ...])	Returns an iterator over all sublayers in the Layer, yielding tuple of name and sublayer.
parameters([include_sublayers])	Returns a list of all Parameters from current layer and its sub-layers.
register_buffer(name, tensor[, persistable])	Registers a tensor as buffer into the layer.
register_forward_post_hook(hook)	Register a forward post-hook for Layer.
register_forward_pre_hook(hook)	Register a forward pre-hook for Layer.
set_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
set_state_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
state_dict([destination, include_sublayers, ...])	Get all parameters and persistable buffers of current layer and its sub-layers.
sublayers([include_self])	Returns a list of sub layers.
to([device, dtype, blocking])	Cast the parameters and buffers of Layer by the give device, dtype and blocking.
to_static_state_dict([destination, ...])	Get all parameters and buffers of current layer and its sub-layers.
train()	Sets this Layer and all its sublayers to training mode.

backward
register_state_dict_hook

forward(x)

conv -> bn -> relu -> conv -> bn + residual connection

class paddlespeech.t2s.models.wavernn.wavernn.UpsampleNetwork(aux_channels: int = 80, upsample_scales: List[int] = [4, 5, 3, 5], compute_dims: int = 128, res_blocks: int = 10, res_out_dims: int = 128, aux_context_window: int = 2)

Bases: Layer

Methods

__call__(*inputs, **kwargs)	Call self as a function.
add_parameter(name, parameter)	Adds a Parameter instance.
add_sublayer(name, sublayer)	Adds a sub Layer instance.
apply(fn)	Applies fn recursively to every sublayer (as returned by .sublayers()) as well as self.
buffers([include_sublayers])	Returns a list of all buffers from current layer and its sub-layers.
children()	Returns an iterator over immediate children layers.
clear_gradients()	Clear the gradients of all parameters for this layer.
create_parameter(shape[, attr, dtype, ...])	Create parameters for this layer.
create_tensor([name, persistable, dtype])	Create Tensor for this layer.
create_variable([name, persistable, dtype])	Create Tensor for this layer.
eval()	Sets this Layer and all its sublayers to evaluation mode.
extra_repr()	Extra representation of this layer, you can have custom implementation of your own layer.
forward(m)	Args:
full_name()	Full name for this layer, composed by name_scope + "/" + MyLayer.__class__.__name__
load_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
named_buffers([prefix, include_sublayers])	Returns an iterator over all buffers in the Layer, yielding tuple of name and Tensor.
named_children()	Returns an iterator over immediate children layers, yielding both the name of the layer as well as the layer itself.
named_parameters([prefix, include_sublayers])	Returns an iterator over all parameters in the Layer, yielding tuple of name and parameter.
named_sublayers([prefix, include_self, ...])	Returns an iterator over all sublayers in the Layer, yielding tuple of name and sublayer.
parameters([include_sublayers])	Returns a list of all Parameters from current layer and its sub-layers.
register_buffer(name, tensor[, persistable])	Registers a tensor as buffer into the layer.
register_forward_post_hook(hook)	Register a forward post-hook for Layer.
register_forward_pre_hook(hook)	Register a forward pre-hook for Layer.
set_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
set_state_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
state_dict([destination, include_sublayers, ...])	Get all parameters and persistable buffers of current layer and its sub-layers.
sublayers([include_self])	Returns a list of sub layers.
to([device, dtype, blocking])	Cast the parameters and buffers of Layer by the give device, dtype and blocking.
to_static_state_dict([destination, ...])	Get all parameters and buffers of current layer and its sub-layers.
train()	Sets this Layer and all its sublayers to training mode.

backward
register_state_dict_hook

forward(m)

Args:

c (Tensor):

Input tensor (B, C_aux, T).

Returns:

Tensor:

Output tensor (B, (T - 2 * pad) * prob(upsample_scales), C_aux).

Tensor:

Output tensor (B, (T - 2 * pad) * prob(upsample_scales), res_out_dims).

class paddlespeech.t2s.models.wavernn.wavernn.WaveRNN(rnn_dims: int = 512, fc_dims: int = 512, bits: int = 9, aux_context_window: int = 2, upsample_scales: List[int] = [4, 5, 3, 5], aux_channels: int = 80, compute_dims: int = 128, res_out_dims: int = 128, res_blocks: int = 10, hop_length: int = 300, sample_rate: int = 24000, mode='RAW', init_type: str = 'xavier_uniform')

Bases: Layer

Methods

__call__(*inputs, **kwargs)	Call self as a function.
add_parameter(name, parameter)	Adds a Parameter instance.
add_sublayer(name, sublayer)	Adds a sub Layer instance.
apply(fn)	Applies fn recursively to every sublayer (as returned by .sublayers()) as well as self.
buffers([include_sublayers])	Returns a list of all buffers from current layer and its sub-layers.
children()	Returns an iterator over immediate children layers.
clear_gradients()	Clear the gradients of all parameters for this layer.
create_parameter(shape[, attr, dtype, ...])	Create parameters for this layer.
create_tensor([name, persistable, dtype])	Create Tensor for this layer.
create_variable([name, persistable, dtype])	Create Tensor for this layer.
eval()	Sets this Layer and all its sublayers to evaluation mode.
extra_repr()	Extra representation of this layer, you can have custom implementation of your own layer.
fold_with_overlap(x, target, overlap)	Fold the tensor with overlap for quick batched inference.
forward(x, c)	Args:
full_name()	Full name for this layer, composed by name_scope + "/" + MyLayer.__class__.__name__
generate(c[, batched, target, overlap, ...])	Args:
load_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
named_buffers([prefix, include_sublayers])	Returns an iterator over all buffers in the Layer, yielding tuple of name and Tensor.
named_children()	Returns an iterator over immediate children layers, yielding both the name of the layer as well as the layer itself.
named_parameters([prefix, include_sublayers])	Returns an iterator over all parameters in the Layer, yielding tuple of name and parameter.
named_sublayers([prefix, include_self, ...])	Returns an iterator over all sublayers in the Layer, yielding tuple of name and sublayer.
pad_tensor(x, pad[, side])	Args:
parameters([include_sublayers])	Returns a list of all Parameters from current layer and its sub-layers.
register_buffer(name, tensor[, persistable])	Registers a tensor as buffer into the layer.
register_forward_post_hook(hook)	Register a forward post-hook for Layer.
register_forward_pre_hook(hook)	Register a forward pre-hook for Layer.
set_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
set_state_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
state_dict([destination, include_sublayers, ...])	Get all parameters and persistable buffers of current layer and its sub-layers.
sublayers([include_self])	Returns a list of sub layers.
to([device, dtype, blocking])	Cast the parameters and buffers of Layer by the give device, dtype and blocking.
to_static_state_dict([destination, ...])	Get all parameters and buffers of current layer and its sub-layers.
train()	Sets this Layer and all its sublayers to training mode.
xfade_and_unfold(y[, target, overlap])	Applies a crossfade and unfolds into a 1d array.

backward
gen_display
progbar
register_state_dict_hook

fold_with_overlap(x, target, overlap)

Fold the tensor with overlap for quick batched inference. Overlap will be used for crossfading in xfade_and_unfold()

Args:

x(Tensor):

Upsampled conditioning features. mels or aux shape=(1, T, features) mels: [1, T, 80] aux: [1, T, 128]

target(int):

Target timesteps for each index of batch

overlap(int):

Timesteps for both xfade and rnn warmup

Returns:

Tensor:

shape=(num_folds, target + 2 * overlap, features) num_flods = (time_seq - overlap) // (target + overlap) mel: [num_folds, target + 2 * overlap, 80] aux: [num_folds, target + 2 * overlap, 128]

Details:

x = [[h1, h2, ... hn]] Where each h is a vector of conditioning features Eg: target=2, overlap=1 with x.size(1)=10

folded = [[h1, h2, h3, h4],

[h4, h5, h6, h7], [h7, h8, h9, h10]]

forward(x, c)

Args:

x (Tensor):

wav sequence, [B, T]

c (Tensor):

mel spectrogram [B, C_aux, T']

T = (T' - 2 * aux_context_window ) * hop_length

Returns:

Tensor: [B, T, n_classes]

gen_display(i, seq_len, b_size, start)

generate(c, batched: bool = True, target: int = 12000, overlap: int = 600, mu_law: bool = True, gen_display: bool = False)

Args:

c(Tensor):

input mels, (T', C_aux)

batched(bool):

generate in batch or not

target(int):

target number of samples to be generated in each batch entry

overlap(int):

number of samples for crossfading between batches

mu_law(bool)

Returns:

wav sequence:

Output (T' * prod(upsample_scales), out_channels, C_out).

pad_tensor(x, pad, side='both')

Args:

x(Tensor):

mel, [1, n_frames, 80]

pad(int):

side(str, optional): (Default value = 'both')

Returns:

Tensor

progbar(i, n, size=16)

xfade_and_unfold(y, target: int = 12000, overlap: int = 600)

Applies a crossfade and unfolds into a 1d array.

Args:

y (Tensor):

Batched sequences of audio samples shape=(num_folds, target + 2 * overlap) dtype=paddle.float32

overlap (int): Timesteps for both xfade and rnn warmup

Returns:

Tensor

audio samples in a 1d array shape=(total_len) dtype=paddle.float32

Details:

y = [[seq1],

[seq2], [seq3]]

Apply a gain envelope at both ends of the sequences

y = [[seq1_in, seq1_target, seq1_out],

[seq2_in, seq2_target, seq2_out], [seq3_in, seq3_target, seq3_out]]

Stagger and add up the groups of samples:

[seq1_in, seq1_target, (seq1_out + seq2_in), seq2_target, ...]

class paddlespeech.t2s.models.wavernn.wavernn.WaveRNNInference(normalizer, wavernn)

Bases: Layer

Methods

__call__(*inputs, **kwargs)	Call self as a function.
add_parameter(name, parameter)	Adds a Parameter instance.
add_sublayer(name, sublayer)	Adds a sub Layer instance.
apply(fn)	Applies fn recursively to every sublayer (as returned by .sublayers()) as well as self.
buffers([include_sublayers])	Returns a list of all buffers from current layer and its sub-layers.
children()	Returns an iterator over immediate children layers.
clear_gradients()	Clear the gradients of all parameters for this layer.
create_parameter(shape[, attr, dtype, ...])	Create parameters for this layer.
create_tensor([name, persistable, dtype])	Create Tensor for this layer.
create_variable([name, persistable, dtype])	Create Tensor for this layer.
eval()	Sets this Layer and all its sublayers to evaluation mode.
extra_repr()	Extra representation of this layer, you can have custom implementation of your own layer.
forward(logmel[, batched, target, overlap, ...])	Defines the computation performed at every call.
full_name()	Full name for this layer, composed by name_scope + "/" + MyLayer.__class__.__name__
load_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
named_buffers([prefix, include_sublayers])	Returns an iterator over all buffers in the Layer, yielding tuple of name and Tensor.
named_children()	Returns an iterator over immediate children layers, yielding both the name of the layer as well as the layer itself.
named_parameters([prefix, include_sublayers])	Returns an iterator over all parameters in the Layer, yielding tuple of name and parameter.
named_sublayers([prefix, include_self, ...])	Returns an iterator over all sublayers in the Layer, yielding tuple of name and sublayer.
parameters([include_sublayers])	Returns a list of all Parameters from current layer and its sub-layers.
register_buffer(name, tensor[, persistable])	Registers a tensor as buffer into the layer.
register_forward_post_hook(hook)	Register a forward post-hook for Layer.
register_forward_pre_hook(hook)	Register a forward pre-hook for Layer.
set_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
set_state_dict(state_dict[, use_structured_name])	Set parameters and persistable buffers from state_dict.
state_dict([destination, include_sublayers, ...])	Get all parameters and persistable buffers of current layer and its sub-layers.
sublayers([include_self])	Returns a list of sub layers.
to([device, dtype, blocking])	Cast the parameters and buffers of Layer by the give device, dtype and blocking.
to_static_state_dict([destination, ...])	Get all parameters and buffers of current layer and its sub-layers.
train()	Sets this Layer and all its sublayers to training mode.

backward
register_state_dict_hook

forward(logmel, batched: bool = True, target: int = 12000, overlap: int = 600, mu_law: bool = True, gen_display: bool = False)

Defines the computation performed at every call. Should be overridden by all subclasses.

Parameters:

*inputs(tuple): unpacked tuple arguments **kwargs(dict): unpacked dict arguments