Recently, I've been trying to make an effort to use my voice recognition software more regularly. It can be fantastic and it's really amazingly accurate in certain contexts, but it does have its drawbacks. Because it's designed primarily for writing continuous text - e-mails, reports, etc - it doesn't always cope very well with the kind of stuff I do as a lexicographer and ELT writer - lots of rather random individual words and phrases, gap-fills, and lots of fiddly formatting of language exercises. It also works best with standard Windows software - it's great for Microsoft Word, but won't work at all with most of the bespoke corpus and lexicography software I use much of the time.
This week though, I've been editing some materials and producing a feedback report in a nice simple Word document, so I've had my headset on for most of the day, sitting back in my chair and chatting away to my computer happily. And overall, it's been pretty successful, a nice break for my hands and generally fairly accurate. It does come up with some odd stuff sometimes though which needs editing by hand. Mostly the errors are fairly boring and a bit annoying, but just occasionally it does come up with a particularly appropriate gem. Yesterday, I dictated "lexicography" and on the screen in front of me appeared "lexical graffiti" - can computer software be responsible for a Freudian slip?! ......
Labels: lexicography, voice recognition software, writing
