If I understand correctly David is saying the lower the frequency the highe=
r the resoulution. For example if the sampling as 44.1 khz if you have a fr=
equency of 10 khz you will have 4 samples of information representing that =
signal - if 22 khz only 2 etc etc it may only be half that come to think of=
it but you get the gist.....
P.S> David - you live in North devon ? You gonna come to Eden ?
--- In "Raimund" <> wrot=
e:
>
> Avocet wrote:
>
> > Theoretically you can sample a 20KHz input at 44.1Ks/s, but it will
> > only have two samples per cycle, so it won't give a sensible 20KHz
> > back. With music, anything up there is hash, but wildlife calls need a=
> > better reproduction than that.
>
> David,
>
> Do you think that it was impossible to perfectly reconstruct such a 20 kH=
z sine signal out of a 44.1 ks/s recording?
>
> I'm afraid that this issue is often poorly understood. Given that the ant=
i-aliasing filter was sharp enough (in such a way that the 20 kHz is not be=
ing attenuated), then it should be possible to perfectly reconstruct the or=
iginal 20 kHz signal and play it back without introducing any distortion. D=
igital oversampling in the D/A converter circuit can help to create a very =
smooth analog output signal that is (almost) identical to the original anal=
og input signal.
>
> Regards,
> Raimund
>
|