espnet.nets.transducer_decoder_interface.TransducerDecoderInterface

source

class espnet.nets.transducer_decoder_interface.TransducerDecoderInterface

Bases: object

Decoder interface for Transducer models.

batch_score(hyps: List[Hypothesis] | List[ExtendedHypothesis], dec_states: Tuple[Tensor, Tensor | None] | List[Tensor | None], cache: Dict[str, Any], use_lm: bool) → Tuple[Tensor, Tuple[Tensor, Tensor | None] | List[Tensor | None], Tensor]

One-step forward hypotheses.

• Parameters:
  • hyps – Hypotheses.
  • dec_states – Decoder hidden states.
  • cache – Pairs of (dec_out, dec_states) for each label sequence. (key)
  • use_lm – Whether to compute label ID sequences for LM.
• Returns: Decoder output sequences. dec_states: Decoder hidden states. lm_labels: Label ID sequences for LM.
• Return type: dec_out

create_batch_states(states: Tuple[Tensor, Tensor | None] | List[Tensor | None], new_states: List[Tuple[Tensor, Tensor | None] | List[Tensor | None]], l_tokens: List[List[int]]) → Tuple[Tensor, Tensor | None] | List[Tensor | None]

Create decoder hidden states.

• Parameters:
  • batch_states – Batch of decoder states
  • l_states – List of decoder states
  • l_tokens – List of token sequences for input batch
• Returns: Batch of decoder states
• Return type: batch_states

init_state(batch_size: int) → Tuple[Tensor, Tensor | None] | List[Tensor | None]

Initialize decoder states.

• Parameters:batch_size – Batch size.
• Returns: Initial decoder hidden states.
• Return type: state

score(hyp: Hypothesis, cache: Dict[str, Any]) → Tuple[Tensor, Tuple[Tensor, Tensor | None] | List[Tensor | None], Tensor]

One-step forward hypothesis.

• Parameters:
  • hyp – Hypothesis.
  • cache – Pairs of (dec_out, dec_state) for each token sequence. (key)
• Returns: Decoder output sequence. new_state: Decoder hidden states. lm_tokens: Label ID for LM.
• Return type: dec_out

select_state(batch_states: Tuple[Tensor, Tensor | None] | List[Tensor], idx: int) → Tuple[Tensor, Tensor | None] | List[Tensor | None]

Get specified ID state from decoder hidden states.

• Parameters:
  • batch_states – Decoder hidden states.
  • idx – State ID to extract.
• Returns: Decoder hidden state for given ID.
• Return type: state_idx