Subject: Re: Frequency response was Re: [linux-audio-user] Audiophile CD's
From: nixx (nixx_AT_nixx.org.uk)
Date: Mon Jan 28 2002 - 20:24:21 EET
On Mon, 28 Jan 2002 17:27:46 -0000
"Tony Lambley" <tonyl_AT_vextech.net> wrote:
> > > What happens when you sample a 1Hz sine with 2 samples a second, and
> those
> > > samples happen to coincide with the zero crossover points?
> >
> > That's why it's necessary to sample at f > 2b -- information is lost
> > when there isn't a sample close enough to the peak/trough of the wave.
> > I'm not sure how much greater than 2b f has to be, but apparently others
> > have determined that 44.1khz or 48khz is high enough to reproduce
> > frequencies up to 20kHz.
>
> But doesn't that still leave a huge potential loss of information? Our ears
> can't tell the difference between a sine and square at 20k, so a 44.1k or
> 48k sample rate is sufficient for a human intended audience.
>
> Wouldn't fx algorithms be a difference case though?
remember that in DSP systems (ie all digital audio systems) analogue
low-pass filters are used to prevent aliasing. low pass filters can never
be 'ideal', so in reality frequencies well below the nyquist rate are
attenuated. (frequencies are only reproduced faithfully up to about 4/5
the nyquist rate if i recall my dsp teaching correctly) so a higher
sampling frequencies help a lot in this respect. also take into account
that doubling the frequency only moves up one octave, so big jumps are
necessary.
-nixx
This archive was generated by hypermail 2b28 : Tue Jan 29 2002 - 03:29:44 EET