Speech-to-text applications are on our phones, part of our cars’ infotainment systems, and have even made their way onto our desktops, but despite their almost-ubiquity many of us use them incorrectly. Kevin Purdy over at Fast Company has a handy guide to mastering voice recognition software, full of a number of obvious suggestions–know what the application can and can’t do; keep the microphone clear; utilize any ambient noise-reduction settings–that many of us nonetheless ignore.
The most interesting bit, though (and I’m writing this as someone who is frequently frustrated by his dictation software) is that speech recognition is meant to work with regular speech patterns, but many of us don’t speak normally while using the software. Instead, we over-enunciate, or speak like a robot … and when the software inevitably gets it wrong, we then over-over-enunciate and speak like a very loud robot.
Everyone Purdy quotes, from the chief technical officer for Dragon Dictate software to a Google rep discussing the search giant’s own speech recognition product says basically the same thing: think about what you’re going to say to reduce stop-starts, stutters and slurred words, but then speak as you normally would; the software will eventually recognize your speech patterns and improve.
Read more at Fast Company.
