As broadcasters’ web sites become more mediarich it would be prohibitively expensive to manually captionall of the videos provided. However, many of these videoshave been clipped from broadcast television and would havebeen captioned at the point of broadcast.The recent FCC ruling requires all broadcasters toprovide closed captions for all ‘straight lift’ video clips thathave been broadcast on television from January 2016. FromJanuary 2017 captions will be required for ‘Montages’which consist of multiple clips, and the requirement tocaption clips from live or near-live television will apply fromJuly 2017.This paper presents a method of automaticallyfinding a match for a video clip from within a set of off-airtelevision recordings. It then shows how the required set ofcaptions can be autonomously identified, retimed andreformatted for use with IP-delivery. It also shows howcaptions can be retrieved for each sub-clip within a montageand combined to create a set of captions. Finally it describeshow, with a modest amount of human intervention, livecaptions can be corrected for errors and timing to provideimproved captions for video clips presented on the web.
展开▼