espnet2.tts2.feats_extract.abs_feats_extract.AbsFeatsExtractDiscrete

source

class espnet2.tts2.feats_extract.abs_feats_extract.AbsFeatsExtractDiscrete(*args, **kwargs)

Bases: Module, ABC

Abstract base class for extracting discrete features from input sequences.

This class provides an interface for subclasses to implement methods for parsing discrete token sequences into structured data formats suitable for prediction tasks. The specific implementations may include options such as retaining the sequence as-is, resizing it into a matrix, or applying multi-resolution techniques.

None

• Parameters:
  • input (torch.Tensor) – A tensor containing the input discrete token sequence.
  • input_lengths (torch.Tensor) – A tensor indicating the lengths of the input sequences.
• Returns: A tuple containing the processed features and a : dictionary with additional information.
• Return type: Tuple[Any, Dict]
• Raises:NotImplementedError – If the forward method is not implemented by a subclass.

####### Examples

>>> model = SomeConcreteFeatsExtractDiscrete()  # An example subclass
>>> input_tensor = torch.tensor([[1, 2, 3], [4, 5, 6]])
>>> input_lengths = torch.tensor([3, 3])
>>> features, info = model(input_tensor, input_lengths)

NOTE

This is an abstract class and cannot be instantiated directly. Subclasses must implement the forward method.

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

abstract forward(input: Tensor, input_lengths: Tensor) → Tuple[Any, Dict]

Forward method for processing a discrete token sequence.

This method takes an input tensor representing a sequence of discrete tokens and their corresponding lengths. It is expected to transform the input into a structured format suitable for further prediction tasks.

• Parameters:
  • input (torch.Tensor) – A tensor containing the input token sequence.
  • input_lengths (torch.Tensor) – A tensor containing the lengths of each input sequence.
• Returns: A tuple where the first element can be any type representing the processed output, and the second element is a dictionary containing additional information related to the processing.
• Return type: Tuple[Any, Dict]
• Raises:NotImplementedError – This method must be overridden by subclasses.

####### Examples

>>> model = SomeConcreteModel()  # A subclass of AbsFeatsExtractDiscrete
>>> input_tensor = torch.tensor([[1, 2, 3], [4, 5, 0]])
>>> lengths = torch.tensor([3, 2])
>>> output, info = model.forward(input_tensor, lengths)