When Google Voice Search was first launched, they wanted to make it work out of the box, without an initial setup process. So Google built speech
models broad enough to accommodate a wide variety of people, regardless
of gender, age, and accents, or variations in pitch, pace, and other
factors.
Google didn't want it to require training. Now, however, there is personalized recognition to train Google Search to learn your personal voice patterns and pitch.
If you opt into personalized recognition, it will begin to associate the
recordings of your words with your Google
account. Google then automatically uses these words to build a speech model
specifically for you. This speech model enables Google to deliver greater
recognition accuracy. Although subtle, accuracy improvements begin
fairly quickly and will build over time.
first time you use Voice Search, you'll be presented with a dialog to
turn on personalized recognition.
personalized recognition functionality is currently available for
English in the United States. Google plans to support other countries and
languages in the near future. To enable it, you must have Android 2.2
or higher, and you must download the latest version of the Voice Search app from
Android Market. The new app also includes some improvements to name
recognition and speed, especially over 3G and EDGE connections.