espnet2.layers.abs_normalize.AbsNormalize

source

class espnet2.layers.abs_normalize.AbsNormalize(*args, **kwargs)

Bases: Module, ABC

Abstract base class for normalization layers in the ESPnet framework.

This class defines the interface for normalization layers that will inherit from it. The derived classes must implement the forward method to specify the normalization behavior. The forward method takes an input tensor and optionally an input lengths tensor, and it returns the normalized output tensor along with its lengths.

None

• Parameters:
  • input (torch.Tensor) – The input tensor to be normalized.
  • input_lengths (torch.Tensor , optional) – The lengths of the input tensor. Defaults to None.
• Returns: A tuple containing the normalized output tensor and its corresponding lengths tensor.
• Return type: Tuple[torch.Tensor, torch.Tensor]
• Raises:
  • NotImplementedError – If the derived class does not implement the forward
  • method. –

####### Examples

class MyNormalize(AbsNormalize): : def forward(self, input: torch.Tensor, input_lengths: torch.Tensor = None): : # Implement normalization logic here return output, output_lengths

normalizer = MyNormalize() output, lengths = normalizer(torch.randn(10, 20), torch.tensor([20] * 10))

Initialize internal Module state, shared by both nn.Module and ScriptModule.

abstract forward(input: Tensor, input_lengths: Tensor | None = None) → Tuple[Tensor, Tensor]

Executes the forward pass of the normalization process.

This method must be implemented by subclasses of AbsNormalize. It takes an input tensor and an optional tensor of input lengths, and returns the normalized output tensor along with the lengths of the output.

• Parameters:
  • input (torch.Tensor) – The input tensor to be normalized.
  • input_lengths (torch.Tensor , optional) – The lengths of the input sequences. Defaults to None.
• Returns: A tuple containing the normalized output : tensor and the corresponding output lengths.
• Return type: Tuple[torch.Tensor, torch.Tensor]
• Raises:NotImplementedError – If this method is called directly without being overridden in a subclass.

####### Examples

>>> model = SomeNormalizationModel()  # Assuming SomeNormalizationModel
>>> input_tensor = torch.randn(10, 5)  # Example input
>>> lengths = torch.tensor([5] * 10)  # Example lengths
>>> output, output_lengths = model.forward(input_tensor, lengths)