espnet.nets.pytorch_backend.fastspeech.duration_predictor.DurationPredictorLoss

source

class espnet.nets.pytorch_backend.fastspeech.duration_predictor.DurationPredictorLoss(offset=1.0, reduction='mean')

Bases: Module

Loss function module for duration predictor.

The loss value is Calculated in log domain to make it Gaussian.

Initilize duration predictor loss module.

• Parameters:
  • offset (float , optional) – Offset value to avoid nan in log domain.
  • reduction (str) – Reduction type in loss calculation.

forward(outputs, targets)

Calculate forward propagation.

• Parameters:
  • outputs (Tensor) – Batch of prediction durations in log domain (B, T)
  • targets (LongTensor) – Batch of groundtruth durations in linear domain (B, T)
• Returns: Mean squared error loss value.
• Return type: Tensor

NOTE

outputs is in log domain but targets is in linear domain.