[music-dsp] OT: CD Perfection

Joshua Scholar joshscholar at yahoo.com
Fri Oct 15 01:22:56 EDT 2004

> The Shannon Nyquist interpolation theory states that any band limited
> signal can be perfectly reconstructed when sampled at the appropiate rate.
> So in theory shouldn't we beable to downsample using a signal which was
> the same as the original analog wave.
> But I agree, in practise this dosn't seem to work. Why is this?

In practice it works perfectly.
I've written up and downsampling programs based on very long linear phase

It's always seemed like magic to me to upsample a sinewave that was sampled
very close to the nyquest and see it change from what looked like an
amplitude modulated wave into a perfect sine wave...

If resampling doesn't work perfectly when you try it then you're using poor
quality software.  High quality software DOES preserve all of the nuances of
the sound, but it does require nontrivial processing.

More information about the music-dsp mailing list