06-16-2012 06:00 AM - edited 06-16-2012 06:01 AM
there are "methods":
- aquire the sound signal
- do some mathematical analysis (like FFT...)
- compare the result with your database of the supported commands...
Maybe you should ask engineers working for Apple or Samsung to tell you how they made Siri or S-Voice. Atleast Apple uses some hardware that supports the mathematical analysis step...