For more than 20 years, audio descriptions ... descriptions via text-to-speech (TTS) methods, rather than using human recordings. IBM and NCAM explored two approaches which exploit new HTML5 media ...