[linux-audio-user] sampling at high frequencies

[Joern Nettingsmeier]

Steve Harris wrote:
> On Mon, Jan 26, 2004 at 10:19:59 +0100, Joern Nettingsmeier wrote:
> 
>>imnsho, all this sampling rate hype is a clever industry ploy to keep 
>>people from thinking that their hardware is finally good enough and they 
>>can stop buying now.
> 
> 
> Probably true but...
>  
> 
>>people can't hear over 20k. period. 48k sampling rate gives you 24k 
>>minus what's cropped off by the aliasing filter. granted, higher 
>>sampling allows you to use a simpler, less steep aliasing filter, and 
>>some people claim to perceive an improvement from that. but even then, 
>>96k should be enough.
> 
> 
> This is not quite correct. While humans can not hear continuous tones over
> 20-odd kHz that is not the whole story, the frequency reproduction also
> limits the minimum transient rise time, which is detected by a different
> part of the ear (IIRC, IANABiologist and my psychoacoustics textbook is at
> work).

interesting point. but isn't membrane inertia the limiting factor in 
transient reproduction anyway?

my naive understanding is that min rise time = nyquist freq. if we 
cannot perceive frequencies above, say, 20k, all we win by faster 
sampling is more accurate timing information.
but there is a worst-case "timing error" of 1/24000 sec, which does not 
seem much to me...
i'd like to read more, but i haven't been able to google anything up 
about transient perception and reproduction. any pointers?

for the record, here's a 1998 aes paper that elaborates on the 
consequences of higher sampling on filter artifaces in the audible 
range: http://www.nanophon.com/audio/antialia.pdf

-- 
"I never use EQ, never, never, never. I previously used to use mic
positioning but I've even given up on that too."
	- Jezar on http://www.audiomelody.com


Jörn Nettingsmeier
Kurfürstenstr 49, 45138 Essen, Germany
http://spunk.dnsalias.org (my server)
http://www.linuxaudiodev.org (Linux Audio Developers)