ConceptSimultaneous Speech Translation / 同声语音翻译
SimulST
A subfield of speech translation in which a system produces target-language output incrementally while the speaker continues to provide input, rather than waiting for a complete utterance. The task requires dynamic trade-offs between translation quality and output latency.
Timeline (last 90 days)
Oct 5
Measuring the performance of SimulST systems typically uses two dimensions: translation quality (BLEU score) and latency (Average Lagging AL, Differentiable AL), which together form the quality–latency frontier curve.
Unverified50%