[music-dsp] A question about Dirac pulses

Glen Berry spamtemp at charter.net
Sat Jan 10 11:03:01 EST 2004

At 01:02 PM 1/10/04 , you wrote:

>A half sample rectangle is down about -1 dB at 22050 Hz and
>a quarter sample about -.24 dB at 22050.  This is all
>without considering the practicalities of oversampled and
>bandwidth limited A/D converters.

So, measured as a percentage of a single sample period, what's the
recommended impulse duration, if I build a device where the impulse time is
controllable? It seems to me that anything between 25% and 50% of a sample
period gives good results. Am I mistaken?

Also, does a longer length pulse actually provide a louder response,
thereby increasing the S/N ratio (definitely a good thing), or does it
mostly reduce the accuracy of the results compared to the shorter pulse.
(This assumes that both short and long pulses are less than one sample
period.) I would think that it would only "muddy" the results if the pulse
were longer than a sample period, but I don't really know for sure.

Glen Berry

More information about the music-dsp mailing list