espnet2.enh.layers.uses.ChannelAttention

source

class espnet2.enh.layers.uses.ChannelAttention(input_dim, att_heads=4, att_dim=256, activation='relu', eps=1e-05)

Bases: Module

Channel Attention module.

• Parameters:
  • input_dim (int) – dimension of the input feature.
  • att_heads (int) – number of attention heads in self-attention.
  • att_dim (int) – projection dimension for query and key before self-attention.
  • activation (str) – non-linear activation function.
  • eps (float) – epsilon for layer normalization.

forward(x, ref_channel=None)

ChannelAttention Forward.

• Parameters:
  • x (torch.Tensor) – input feature (batch, C, N, freq, time)
  • ref_channel (None or int) – index of the reference channel.
• Returns: output feature (batch, C, N, freq, time)
• Return type: output (torch.Tensor)