"The reason is VERY simple. The listening for the "OK Google" phrase is local, it is done on the device only, NOTHING is sent to Google until the device hears the built in phrase. The phrase was chosen to provide the least amount of false positives. If people can willy nilly change the phrase the amount of false positives will skyrocket and people will very quickly get frustrated with the tech. Speech recognition with cloud processing is amazingly accurate, speech recognition via a low power chip has not changed for years - especially when you think that it has to cope with accents too."
TomTom uses "Hello TomTom", locally processed too, and that occasionally gets tricked by things said on the radio. Quite annoying really because now I have the radio and the TomTom talking over each other and the TomTom won't understand "cancel" until the radio is switched off.
Otherwise it's pretty good. It doesn't use the cloud at all for its voice recognition, it's all done locally. It has a limited vocabulary, but I can say addresses quite a few commands to it and that (mostly) works. Certain it's pretty good for a small device with no online assistance from Nuance or whoever. Moore's law means that one day it'll all be locally processed which will be useful.
The only problem with "Hello TomTom" is that the name feels like it should be masculine. And then "Jane" (the default voice) calmly speaks back to you. Now I'm sure I'm not alone in finding Jane's voice alluring, but to discover that she’s a tomboy too, well that knocks one's concentration completely skew wiff.
(anon in case missus is reading)
Biting the hand that feeds IT
