[sdiy] Has anyone built a voice pitch tracker?

[Buck Buchanan]

Dan wrote:

> 1. I need to eliminate the second harmonic - and when I sing very low 
> the second harmonic is within the range of interest, so it can´t be 
> filtered out with conventional filters.


Ah, the joys of pitch extraction.  I've had exactly the same issue (I 
think) when trying to extract pitch from an electric guitar signal.

While I don't have a specific answer for you, in Electronotes # 53 
Bernie Hutchins covers a technique which might be worth experimenting with.

It basically involves using a peak detector type envelope follower to 
generate a DC voltage that tracks the peaks of the original waveform. 
The time constant of this should be adjusted so it doesn't miss too many 
peaks (it will have lots of ripple but I think that's ok).

Then take this envelope (properly scaled), and use it as a reference to 
a comparator of which the original signal is the input.  The idea is to 
trigger the comparator on the fundamental peaks and not the second 
harmonic peaks.  In a working situation, the output should be a pulse 
train at fundamental frequency.

This technique falls apart when the second harmonic is as big as the 
fundamental (this happened in my guitar example and pre-filtering plus 
some other jumping through of hoops helped).

I'm not sure if that helps you get where you're going but I thought I'd 
throw it out there.  I'd be happy to copy the article for you.  Let us 
know what you come up with!

Buck