Friday, 28 March 2014

Voice Recognition Software and I: Part 1

After I did the incredibly brave (read stupid) thing to write 100,000 words during NaNoWriMo in 2012, I vowed to never to do that again. My hands hurt for weeks afterwards. Given the fact that I am responsible for writing a large report to our project's client every month, it had not been a smart thing to do.
Since then I have been looking out for voice recognition software that was easy to use, had a good accuracy rating, was compatible with my accent (I speak English, but with a South African accent) and was affordable.
The last requirement had been the most difficult. The one package, Dragon Naturally Speaking, fit all my needs, except for the aforementioned price. I found it on Amazon, but had to wait quite a while for a special offer to come around to be able to afford the package, the shipping and duties to South Africa.

Using the software
  1. Take the time to let the software "learn" your voice. Every person is unique and the software can only work with a good measure of accuracy if you take the time.
  2. DNS also learns your writing style from books, emails, articles, etc. that you have already written. Be patient while it works through these files.
  3. Pronunciation needs to be clear. Computers do not understand mumbling and incoherent speech, although people do.
  4. Invest in a gaming headset with microphone. These kinds of headsets are of higher quality and aimed at interaction with a computer.

Part 2 will be my personal experience with using DNS software, and other comments on writing fiction with voice recognition software :)