For their final project, a group of students at cornell university decited to make a smart voice decoder for vowels – this basically acquires samples of your voice, do the wave math and determines which vowels are being spoken.
The input is captured using a microphone through an amplifier circuit. Voice data is analyzed in real time (ill say it again REAL TIME) using a Mega644 microcontroller – it has both hardware and soft ware intrefaces that would trigger the analysis – the user could use a button (yellow) or could enter a command in PuTTY.
As an application of their methods they also built a simple voice password system. Where the only way to gain access (unlock) is to correctly utter a sequence of vowels at the correct order. I really like this project specially that it touches the topics of Fast Walsh Transform, sampling theorems and human speech analysis. We are waiting for the version 2 of this to come out, come on people add the consonants ide like to try “Open Sesame” on our house!