ENCO has added automatic language detection to enCaption for live programmes that move between speakers and languages. The system can recognise changes across combinations of 50 supported languages and generate captions without an operator manually changing the active language.
The Linux-based enCaption Sierra platform now integrates enTranslate for real-time on-premises text translation, keeping content inside the facility. The new enSpeak layer converts translated text into low-latency synthetic speech for secondary audio, streams or personal devices.
Together, the components form a speech-to-caption-to-translation-to-voice chain. That can reduce manual switching in multilingual news, panels and sport while giving broadcasters a local-processing option for sensitive live content.
ENCO has not published language-by-language accuracy, diarisation performance or end-to-end delay. Broadcasters need monitored fallback paths, human escalation and clear disclosure because an error can propagate from recognition into both translated captions and spoken output.
