Improving access to Radio and Television for Deaf and Hard of Hearing people in Botswana Through Free Automatic Speech Recognition Technology

Abstract

The aim of the study was to propose the Automatic Speech Recognition (ASR) technology as a means of improving access to televised and radio broadcasted government communications in English language for people who are deaf and hard of  hearing (DHH) in Botswana. To gain better insight into possibilities of using ASR for information accessibility for people who are DHH, this paper evaluated the accuracy rate of the ASR technology in terms of Word Error Rate (WER) by comparing transcriptions generated by ASR software and transcriptions generated and edited by the researcher. The results of our study showed that all the audio-visual documents transcribed scored a WER of less than 15%, indicating that transcriptions were of a very high quality. Some documents even scored an impressive WER of less than 5%, meaning that the transcriptions generated could be used and understood without any human intervention. The study concludes that freely available ASR technology is adapted as a means of improving and diversifying accessibility modes to televised and radio broadcasted government communications in English language for people who are DHH in Botswana.