espnet.tts package

Initialize sub package.

espnet.tts.__init__

Initialize sub package.

espnet.tts.pytorch_backend.tts

E2E-TTS training / decoding functions.

class espnet.tts.pytorch_backend.tts.CustomConverter[source]

Bases: object

Custom converter.

Initilize module.

class espnet.tts.pytorch_backend.tts.CustomEvaluator(model, iterator, target, device)[source]

Bases: espnet.utils.training.evaluator.BaseEvaluator

Custom evaluator.

Initilize module.

Parameters
  • model (torch.nn.Module) – Pytorch model instance.

  • iterator (chainer.dataset.Iterator) – Iterator for validation.

  • target (chainer.Chain) – Dummy chain instance.

  • device (torch.device) – The device to be used in evaluation.

evaluate()[source]

Evaluate over validation iterator.

class espnet.tts.pytorch_backend.tts.CustomUpdater(model, grad_clip, iterator, optimizer, device, accum_grad=1)[source]

Bases: chainer.training.updaters.standard_updater.StandardUpdater

Custom updater.

Initilize module.

Parameters
  • model (torch.nn.Module) – Pytorch model instance.

  • grad_clip (float) – The gradient clipping value.

  • iterator (chainer.dataset.Iterator) – Iterator for training.

  • optimizer (torch.optim.Optimizer) – Pytorch optimizer instance.

  • device (torch.device) – The device to be used in training.

update()[source]

Run update function.

update_core()[source]

Update model one step.

espnet.tts.pytorch_backend.tts.decode(args)[source]

Decode with E2E-TTS model.

espnet.tts.pytorch_backend.tts.train(args)[source]

Train E2E-TTS model.

espnet.tts.pytorch_backend.__init__

Initialize sub package.