Subject: Re: [linux-audio-dev] Sample Rate / Resolution question
From: Paul Winkler (pw_lists_AT_slinkp.com)
Date: Thu Feb 28 2002 - 23:18:15 EET
On Thu, Feb 28, 2002 at 08:54:09PM +0100, Pieter Palmers wrote:
>
>
> >My question is on Sample Rate. If I am going to be putting something on
> >CD eventually, is it better to work with 44.1 kHz or is it more beneficial
> >to work at the higher rates? The reason I ask is 96 Khz does not divide
> >equally by 44.1 kHz, so I was wondering if you would get aliasing effects,
> >since its not equal, and if so, are their benefits in quality from the
> >higher rate that outweigh this?
>
> it doesn't matter if 96 & 44.1 divide equally or not. you will always have
> to do
> anti-alias filtering before downsampling.
This is true, but it does matter for another reason: resampling to a non-integral
ratio requires more sophisticated interpolation and is less likely to be accurate.
So for work targeted to CD, 88.2 would be a better sampling rate, assuming
you believe it's any better than 44.1. I haven't yet convinced myself it's necessary.
> the aliasing effect always appears when downsampling a signal with frequency
> content above the (target sampling frequency)/2. You just have to get rid of
> any frequency content above 22050Hz before downsampling to 44000hz.
>
> So why do we work with 96kHz? I don't know of any reason other than 'it
> seems to
> sound better'.
and it's a convenient excuse for the industry to sell us more expensive gear...
--PW
This archive was generated by hypermail 2b28 : Thu Feb 28 2002 - 23:12:08 EET