Inheritance diagram for model.Attention:

Collaboration diagram for model.Attention:

Public Member Functions
def	__init__ (self, attention_rnn_dim, embedding_dim, attention_dim, attention_location_n_filters, attention_location_kernel_size)

def	get_alignment_energies (self, query, processed_memory, attention_weights_cat)

def	forward (self, attention_hidden_state, memory, processed_memory, attention_weights_cat, mask)

Public Attributes
	query_layer

	memory_layer

	v

	location_layer

	score_mask_value

Constructor & Destructor Documentation

◆ init()

def model.Attention.__init__	(	self,
		attention_rnn_dim,
		embedding_dim,
		attention_dim,
		attention_location_n_filters,
		attention_location_kernel_size
	)

Member Function Documentation

◆ get_alignment_energies()

def model.Attention.get_alignment_energies	(	self,
		query,
		processed_memory,
		attention_weights_cat
	)

PARAMS
------
query: decoder output (batch, n_mel_channels * n_frames_per_step)
processed_memory: processed encoder outputs (B, T_in, attention_dim)
attention_weights_cat: cumulative and prev. att weights (B, 2, max_time)

RETURNS
-------
alignment (batch, max_time)

Here is the caller graph for this function:

◆ forward()

def model.Attention.forward	(	self,
		attention_hidden_state,
		memory,
		processed_memory,
		attention_weights_cat,
		mask
	)

PARAMS
------
attention_hidden_state: attention rnn last output
memory: encoder outputs
processed_memory: processed encoder outputs
attention_weights_cat: previous and cummulative attention weights
mask: binary mask for padded data

Here is the call graph for this function:

Member Data Documentation

◆ query_layer

model.Attention.query_layer

◆ memory_layer

model.Attention.memory_layer

◆ v

model.Attention.v

◆ location_layer

model.Attention.location_layer

◆ score_mask_value

model.Attention.score_mask_value

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

demo/Tacotron2/tacotron2/model.py

Public Member Functions

Public Attributes