Obvious accents or impediments aside, everyone’s speech is a little bit different. I, for example, tend to not pronounce the “t” in “exactly”; as a result, Google Voice Search will occasionally assume I’m instead saying the incredibly common phrase, “eggs acting”.
Eggs, act no more! This morning, Google is announcing an optional “Personalized Recognition” mode for Voice Search on Android.
By analyzing the little snippets of audio you send its way, Voice Search will eventually be “trained” to know all those little quirks of your speech patterns*.
Before you plop on the tin-foil hat and swear to never use Google Voice Search again, let me reiterate: this is an opt-in feature. Once you’ve checked the box to opt-in, recordings of the words you’re voice-searching for will be associated with your Google account — that way, your training follows you to any Android device (or, be it that they open up this feature to other platforms, any application tied to a Google account). Realize after the fact that your twice-daily voice searches for Hello Kitty fan fiction might seem a bit strange out of context? You can clear your voice recordings in your Google Dashboard.
As is par for the course when it comes to voice-recognition stuff, the availability is a bit limited at first. You’ll need Android 2.2 or higher on your handset, you’ll need to be in the U.S., and you’ll need to be an English speaker. Does that describe you? Pop into the Android Market and grab the latest Voice Search app, and get to training!
(* Plus, this way the robot clone of you that they build in 2127 will totally sound just like you)