Abstract for group Pitch

We have decided to create a real-time pitch detection system with
graphical display. It is based on finding the first harmonic of the
human voice for small increments of time, and then plotting it
live(real time) in the time domain. The main bottleneck will be the
autocorrelation process, and we will need to determine, based on
available memory and processor speed, the feasability of our
current algorithm which can be viewed at:

www.columbia.edu/~gs2131/tonedetection/

*View algorithm page.

Other than being a cool toy, this will help people learn about the
pitch of the voice coming out of their mouth. More specifically,
this will allow beginner language students to learn chinese or
other pitch/pitch change dependant languages more easily. By
comparing their own pitch patterns with the pitch patterns of a
fluent speaker, the user will be able to vary their own pitch
patterns in real time to the correct pitch required.

Our preliminary plan is to use a microphone at the input, run the
resulting analog signal through a Low Pass Filter, convert the
analog signal into digital, process the information using a version
of the algorithm shown earlier, and then finally display the real
time plot on the computer.