As I promised myself to dig more about Chrome’s Voice recognition, I have started it already after having a cup of tea and Honey Nut Cherios. By the way I found them very tasty.
Coming back to Chrome, first thing I found is their code trunk :
Here is the link to speech recognition speech_recognizer.cc
Mugging it more I came to know they are using FLAC – Free Lossless Audio Codec, here is the link
Checking more about FLAC I came to know it compress audio by 50% but the good thing is it does not lose a single bit during the process.
If you want to test Chrome Voice Recognition, checkout this out
Then I scratched my itching and started to search for a site where I can find some phonetic sounds that I used on HTML5Rocks.
Here is the Link
I used Phonetics in American English, Spanish and German
I tried few words like ball, taboo, rub , goose, ago, dear and much more .. here are the results
ball – lol
taboo – taboo
rub – various attempts give krupp, growth,
goose – who is
ago – google and piccola (God what is this?)
dear – dear
Results are not impressive !!! Damn………….. !
My sincere advice to them, please please please do a sincere alpha testing… get people on board from various countries and see if they can extend support for
various phonetics.. BEFORE that see what improvements are required in the original APIs.