 |
 |
>Does anyone know if any of this is robust enough to convert taped speech to >text - i.e. taped lectures? > Not a hope! - Well, not very likely anyway. Obviously a discrete speech system won't work at all (unless you can persuade the lecturers to...start...giving....their...lectures....like.... this!) Continuous speech systems, maybe, but you'd need very high quality recordings. Perhaps using a lapel mic with the recorder in the lecturer's pocket or an FM radio mic. You'd also need to get the lecturer to train up the system before you start. Personally, I think that it's unlikely to work except in the most formally presented lectures. Any umming and erring or just poorly enunciated speech will wreck the process. This is a bit of an aside but related.... Does anyone know how they do the real-time Teletext subtitling on the news? I've always believed this was done by a very high power speech system (the Phillip's one perhaps?) because of the type of mistakes you see. Is there some human intervention going on at the same time to fix those errors? Phil Satchell Technical Project Officer, Office for Students with Disabilities Open University > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%