[music-dsp] analogue warmth?

shifty at sidehack.sat.gweep.net shifty at sidehack.sat.gweep.net
Mon Jan 15 16:51:11 EST 2001

Is it true that simply sampling things near the Nyquist rate
results in distortion?

e.g. when you sample a 22.0KHz sine wave at 44.1KHz, you see
a 100Hz modulation.  What causes that?  Would it be fixed with
an ideal brickwall filter?  

Also, is true there are two levels of CRC on CD's?  Why, if they're
so perfect?  is it possible that a lot more errors go on than
you might expect?  Also, does aluminum deteriorate faster or
slower than golden audio CD's?

Finally, where can we see a graph of the phase response of a typical
CD-audio recovery filter?


> >On[ Mon, 15 Jan 2001, Martin Saleteg wrote:
> >
> >> Well, I for one am a firm believer that 16-bit/44.1kHz gives a very bad
> >> representation of frequencies between say Nyquist/2 and Nyquist. 
> >
> >I'm a firm believer that for final ready-for-listening sound storage,
> >anything above 44.1kHz is as unnecessary as pressed golden audio CDs
> >are.
> There's some research that shows that, while that would be true in theory, 
> filter design is hard enough that 96kHz, or even just 48 kHz, gives you better 
> sound out of commercially available products. The reason for this is NOT 
> that we hear stuff over 22 kHz (we don't) but the fact that a brick wall 
> cutoff filter at 20 kHz with Nyquist at 22 kHz does really nasty things to 
> the signal, whereas the same filter with Nyquist at 24 kHz only needs HALF 
> the rejection rate, and at 48 kHz needs ONE FOURTEENTH the rejection 
> rate (steepness).
> Also, passband ripple from these filters is suggested as one contributing 
> factor to the inadequacy of current 44 kHz based players.
> >> Describing a 15kHz sinewave with only 44100 samples per second doesn't
> >> give a very accurate representation of the original signal. 
> >
> >It gives an equally good representation of a 15kHz sinewave as it does of
> >a 100Hz sinewave, so it sounds like we are talking about a bad
> The problem is that we very seldom enjoy recordings of static sine waves. 
> The transients, glissandos and other variations found in music (live music 
> especially) can easily put spectral components of that 15 kHz signal towards 
> the Nyquist limit.
> And, indeed, a static sine wave is not the single frequency spike you'd think; 
> if that's what you want, you'd need to listen to a sin(x)/x signal...
> Cheers,
> 				/ h+


dupswapdrop -- the music-dsp mailing list and website: subscription info,
FAQ, source code archive, list archive, book reviews, dsp links

More information about the music-dsp mailing list