espnet3.systems.asr.metrics.wer.WER

source

class espnet3.systems.asr.metrics.wer.WER(ref_key: str = 'ref', hyp_key: str = 'hyp', clean_types: Iterable[str] | None = None)

Bases: AbsMetric

Compute WER for hypotheses.

This metric expects hypothesis and reference strings and produces a percentage score along with alignment visualization output.

Initialize the WER metric.

• Parameters:
  • ref_key – Key name for reference text entries.
  • hyp_key – Key name for hypothesis text entries.
  • clean_types – Optional cleaner types passed to TextCleaner.