Re: Frequencies was Re: AW: AW: OT: new Macbook wíthOUT Firewire :(

[van Sinn <vansinn@post.cybercity.dk>]

Uh oh, I find it hard arguing with an authority like you ;)


Bob Amstadt wrote:
> I haven't done a research study on sample rates, but I have studied the 
> subject in the classroom and in the real world.  As you increase the 
> sample rate, you build a better quality representation of the original 
> signal.  In terms of what you hear, this will translate to a lower noise 
> floor. Personally, I don't know how much better, but if you have 
> equipment capable of handling 192kHz sampling instead of 48kHz, then you 
> might as well use it. If you are a professional recording studio, you 
> want to use the absolute best that you can afford.

Not fully correct.  That the higher sample rate must yield a more 
fine-grained representation of the souce is true, but what matters is 
being able to reproduce the source, which can be done as soon as the 
sample rate match/exeed the minimum as required by Shannon/Nyquist + of 
cause practical considerations WRT oversampling and filtering et al..
It can be shown by math, which I can understand - just don't ask me to 
provide such math proof :)

> On the other hand, you can record at 44.1kHz and 48kHz and find that the 
> quality of sound is exactly what you want.  In fact, if you are going to 
> distribute the music digitally on CD or compatible with a CD then you 
> will be providing the listener with music sampled at 44.1kHz no matter 
> what frequency you originally sampled it at.  There is no advantage to 
> 192kHz sampling and down converting to 48 over simply sampling at 48kHz 
> in the first place.

Agreed.  In fact, I've heard numerous stories about pro audio engineers 
working in 44.1/48 and simply upsampling to 96/192 if clients require 
this high sample rate.  When getting the work back from the client, they 
downsample again.  With correctly applied up/down sampling, dithering 
and noiseshaping, the result is reportedly indistinquisable.

> As for aliasing effects that were discussed, all A/D converters use an 
> anti-aliasing filter before the actual conversion.  This prevents beat 
> frequencies from appearing in the audio.  In fact, modern A/D converters 
> use a two-step anti-aliasing filter that provides a very sharp cut off 
> of the dangerous frequencies.  So, you aren't going to hear beat 
> frequencies in the result because the difference in frequency between 
> the original signal and the sampling rate are large enough that the 
> resulting aliases are outside of the audible rangle.
> 
> Bob

Absolutely true.

-- 
rgds,
van Sinn