[music-dsp] Re: recosntruction from zero crossings

Crackpot shifty at sidehack.sat.gweep.net
Tue Jan 7 08:32:01 EST 2003


On Tue, Jan 07, 2003 at 09:54:06AM +0200, Yaakov Stein wrote:
> > Just an aside, but how do the zero crossings translate into 
> > instantaneous frequency?  I have never thought of it in exactly that way.
> 
> Assuming that the signal has a high enough frequency,
> just counting the number of zero crossings in a (small enough) unit time
> will give (twice) the instantaneous frequency (of course you really should
> start with large time intervals and do the limit dt -> 0).
> 
> If the frequency is too low for that to give accurate estimates,
> you have to do an interpolation. There are several good interpolation
> techniques that rely on only three or five points.

I have done this in real-time on the ez-kit lite (21xx).
audio examples: http://www.gweep.net/~shifty/death/domain/

It sounds neat, squishy.  Accuracy-wise, man is it BAD!
Superposition of multiple sine waves, for example, render it instantly
perplexed.

Take using fixed window sizes,
even a fixed frequency sinusoid will alternate between two values.

other interesting artifacts occur at nominal silent input which
actually contains high frequency noise, if you neglect to include 
an amplitude envelope follower or a squelch control.

But don't let this discourage you...once you have a signal converted
into the Modulation Domain, there are many interesting transformations
you can take it through.

-N

-- 
                                 DSP Audio Effects! http://gweep.net/~shifty
     .        .       .      .     .    .   .  . ... .  .   .    .     .      .
"La la la laaa laaa laaa                   "      |     Niente 
 La la la laaa laaa laaa."  -Stereolab            | shifty at gweep.net



More information about the music-dsp mailing list