This is a public-interest archive. Personal data is pseudonymized and retained under GDPR Article 89.

Re: voice recognition software

There are two software packages that are the most popular one is by IBM (Via Voice) but I don't recall who now owns the second one (Naturally Speaking).  The recent reviews of both indicate around 95% accuracy after 'training' the software.  So, if you get 250 words to a page that means if you're very lucky and have a very easily recognized speech pattern and alliteration you'll get about 13 errors per page.  That assumes nontechnical jargon.  Scientific terms, plant names and the like will probably add errors until the software is completely trained.  Other things that can effect error rates are changes in rooms, background noise, having a cold or sore throat.  The last time I tried the software was about two years ago and I was very disappointed.  What I've read recently hasn't changed my mind.  If you're fingers work, it's still the best rout e.  If however, you are disabled and must rely on your voice, then the software can obviously be a godsend.  The software seems most suited to those in the legal and medical professions...where a great deal of dictation takes place and where the terminology is available as add ons to the software...but these professional packages can be very expensive.  Older computers with processors below the Pentium 3 class are also much too slow.  A Pentium 4 at 2ghz. would be ideal.

My recommendation, see if you can try it first, or at least return it.  In most cases you can't though.  Wish the news was better cause I could get alot of writing done when I do my five hour drive once a week.

Andrew Messinger
The Hampton Gardener
The Hampton Gardener is a Registered Trade Mark
(Published every Thursday in the Southampton Press)

Other Mailing lists | Author Index | Date Index | Subject Index | Thread Index