Quote:
Originally Posted by chbeer
I think google does it on the server. They send the recorded sound to their server and he recognizes it.
There is no hardware supported voice recognition. Not on the iMac nor on the iPhone. You can build your own but the iPhone has "only" a 500 MHz ARM processor... so not much processing power.
|
I believe that Google's service is probably doing something similar to Shazam, and sending either a small token or three generated from the audio locally on the iPhone, or a heavily compressed audio stream (like mono 32Kbps or something awful). If they didn't it would be painfull slow to upload decent recorded quality PCM of your voice from an iPhone to a server for parsing.
