The Effective Affordable Reusable Speech-To-Text (EARS) program is developing speech-to-text (automatic transcription) technology whose output is substantially richer and much more accurate than currently possible. This will make it possible for machines to do a much better job of detecting, extracting, summarizing, and translating important information. It will also enable humans to understand what was said by reading transcripts instead of listening to audio signals.
EARS is focusing on natural, unconstrained human-human speech from broadcasts and telephone conversations in multiple languages. The intent is to create core enabling technology suitable for a wide range of advanced applications.
EARS encompasses wide-ranging, multidisciplinary research; quantitative evaluations of algorithm accuracy and utility; and efficient technology demonstration prototypes. Outside groups are invited to participate in the annual Rich Transcription evaluations run by NIST.