About

Live television programmes such as the news, sports programmes, chat shows and debates are increasingly subtitled by respeaking. With speech rates too high to capture on a traditional keyboard and the training for stenographers expensive and lengthy, respeaking or the process of using speech recognition software to produce the subtitles for a programme as it is broadcast, has become a popular and viable option. It enables the deaf and hard of hearing population, approximately 1 in 6 people in the UK, to access these programmes, and is also available for others to use, amongst whom language learners, in particular, benefit.  My interest lies in taking my experience and training in television subtitling and applying it to unscripted live events, making them accessible to a wider audience. Such events occur across many settings and whilst accessible provision is increasing, accessing the spoken word remains a barrier for many. However, true accessibility will only be achieved if the subtitles are of a good quality, allowing the audience to follow without having to decipher the errors that respeaking through speech recognition software can cause.  In order to achieve this degree of accessibility, four key areas must be considered: firstly, the needs and desires of the deaf, hard of hearing and language learning audiences; secondly, how a respeaker can achieve the optimum balance between the content that can be respoken and accuracy that is achieved, thirdly, how this knowledge can best be shared through training; finally, the steps that can be taken to make the unknown of live events known to the respeaker and recognisable to the speech recognition software to pre-empt situations that would otherwise lead to errors or necessitate paraphrasing and loss of content.  It is hoped that this study will lead to guidelines and training which could be adapted across languages .