espnet3.systems.asr.metrics.ter.TER

source

class espnet3.systems.asr.metrics.ter.TER(bpemodel: str | Path, ref_key: str = 'ref', hyp_key: str = 'hyp', clean_types: Iterable[str] | None = None)

Bases: BaseMetric

Compute TER (token error rate) for a dataset.

TER is the error rate over the model’s subword (BPE) tokens: the reference and hypothesis text are tokenized with a SentencePiece model, then scored like WER over the resulting token sequences. This mirrors espnet2’s Stage 13 scoring, which computes ter at the bpe token level.

Initialize the TER metric.

• Parameters:
  • bpemodel – Path to the SentencePiece model used to tokenize text into subword tokens (typically the recipe’s trained bpe.model).
  • ref_key – Key name for reference text entries.
  • hyp_key – Key name for hypothesis text entries.
  • clean_types – Optional cleaner types passed to TextCleaner.