Translation
Producing captions in a different language from the speaker's source language — typically built on top of source captions in modern captioning systems.
Live caption translation produces captions in target languages other than the source spoken language. The standard pipeline is: source-language ASR produces source captions, a translation engine renders them into target languages, the multi-language stream is distributed to viewers who pick their preferred language.
Translation latency is typically sub-second on top of source-caption latency, so the experience for non-source-language attendees is functionally identical to native speakers reading source captions.
Translation accuracy is bounded by source-caption accuracy and by the language-pair coverage of the underlying translation engine. Major language pairs (English-Spanish, English-French, English-Mandarin) produce excellent translation; smaller-coverage pairs are noisier.