model.Decoder Class Reference

Inheritance diagram for model.Decoder:

Collaboration diagram for model.Decoder:

Public Member Functions
def	__init__ (self, n_mel_channels, n_frames_per_step, encoder_embedding_dim, attention_dim, attention_location_n_filters, attention_location_kernel_size, attention_rnn_dim, decoder_rnn_dim, prenet_dim, max_decoder_steps, gate_threshold, p_attention_dropout, p_decoder_dropout, early_stopping)
def	get_go_frame (self, memory)
def	initialize_decoder_states (self, memory)
def	parse_decoder_inputs (self, decoder_inputs)
def	parse_decoder_outputs (self, mel_outputs, gate_outputs, alignments)
def	decode (self, decoder_input, attention_hidden, attention_cell, decoder_hidden, decoder_cell, attention_weights, attention_weights_cum, attention_context, memory, processed_memory, mask)
def	forward (self, memory, decoder_inputs, memory_lengths)
def	infer (self, memory, memory_lengths)

Public Attributes
	n_mel_channels
	n_frames_per_step
	encoder_embedding_dim
	attention_rnn_dim
	decoder_rnn_dim
	prenet_dim
	max_decoder_steps
	gate_threshold
	p_attention_dropout
	p_decoder_dropout
	early_stopping
	prenet
	attention_rnn
	attention_layer
	decoder_rnn
	linear_projection
	gate_layer

Constructor & Destructor Documentation

__init__()

def model.Decoder.__init__	(		self,
			n_mel_channels,
			n_frames_per_step,
			encoder_embedding_dim,
			attention_dim,
			attention_location_n_filters,
			attention_location_kernel_size,
			attention_rnn_dim,
			decoder_rnn_dim,
			prenet_dim,
			max_decoder_steps,
			gate_threshold,
			p_attention_dropout,
			p_decoder_dropout,
			early_stopping
	)

Member Function Documentation

get_go_frame()

def model.Decoder.get_go_frame	(		self,
			memory
	)

Gets all zeros frames to use as first decoder input
PARAMS
------
memory: decoder outputs

RETURNS
-------
decoder_input: all zeros frames

Here is the caller graph for this function:

initialize_decoder_states()

def model.Decoder.initialize_decoder_states	(		self,
			memory
	)

Initializes attention rnn states, decoder rnn states, attention
weights, attention cumulative weights, attention context, stores memory
and stores processed memory
PARAMS
------
memory: Encoder outputs
mask: Mask for padded data if training, expects None for inference

Here is the caller graph for this function:

parse_decoder_inputs()

def model.Decoder.parse_decoder_inputs	(		self,
			decoder_inputs
	)

Prepares decoder inputs, i.e. mel outputs
PARAMS
------
decoder_inputs: inputs used for teacher-forced training, i.e. mel-specs

RETURNS
-------
inputs: processed decoder inputs

Here is the caller graph for this function:

parse_decoder_outputs()

def model.Decoder.parse_decoder_outputs	(		self,
			mel_outputs,
			gate_outputs,
			alignments
	)

Prepares decoder outputs for output
PARAMS
------
mel_outputs:
gate_outputs: gate output energies
alignments:

RETURNS
-------
mel_outputs:
gate_outpust: gate output energies
alignments:

Here is the caller graph for this function:

decode()

def model.Decoder.decode	(		self,
			decoder_input,
			attention_hidden,
			attention_cell,
			decoder_hidden,
			decoder_cell,
			attention_weights,
			attention_weights_cum,
			attention_context,
			memory,
			processed_memory,
			mask
	)

Decoder step using stored states, attention and memory
PARAMS
------
decoder_input: previous mel output

RETURNS
-------
mel_output:
gate_output: gate output energies
attention_weights:

Here is the caller graph for this function:

forward()

def model.Decoder.forward	(		self,
			memory,
			decoder_inputs,
			memory_lengths
	)

Decoder forward pass for training
PARAMS
------
memory: Encoder outputs
decoder_inputs: Decoder inputs for teacher forcing. i.e. mel-specs
memory_lengths: Encoder output lengths for attention masking.

RETURNS
-------
mel_outputs: mel outputs from the decoder
gate_outputs: gate outputs from the decoder
alignments: sequence of attention weights from the decoder

Here is the call graph for this function:

infer()

def model.Decoder.infer	(		self,
			memory,
			memory_lengths
	)

Decoder inference
PARAMS
------
memory: Encoder outputs

RETURNS
-------
mel_outputs: mel outputs from the decoder
gate_outputs: gate outputs from the decoder
alignments: sequence of attention weights from the decoder

Here is the call graph for this function:

Member Data Documentation

n_mel_channels

model.Decoder.n_mel_channels

n_frames_per_step

model.Decoder.n_frames_per_step

encoder_embedding_dim

model.Decoder.encoder_embedding_dim

attention_rnn_dim

model.Decoder.attention_rnn_dim

decoder_rnn_dim

model.Decoder.decoder_rnn_dim

prenet_dim

model.Decoder.prenet_dim

max_decoder_steps

model.Decoder.max_decoder_steps

gate_threshold

model.Decoder.gate_threshold

p_attention_dropout

model.Decoder.p_attention_dropout

p_decoder_dropout

model.Decoder.p_decoder_dropout

early_stopping

model.Decoder.early_stopping

prenet

model.Decoder.prenet

attention_rnn

model.Decoder.attention_rnn

attention_layer

model.Decoder.attention_layer

decoder_rnn

model.Decoder.decoder_rnn

linear_projection

model.Decoder.linear_projection

gate_layer

model.Decoder.gate_layer

---

The documentation for this class was generated from the following file:

• demo/Tacotron2/tacotron2/model.py