In article , Don Pearce
wrote:
I use Adobe Audition and 88.2k/24 instead, but the principle is the
same. Audition has a sizeable licence fee, but I inherited a copy when
a project was canned.

88.2k not 96k as they are going to CDs, and I think the down-sampling
must be simpler.

You'd think it might be simpler, but in fact the same algorithm is used.
But this is something Audition (inherited from its life as CoolEdit Pro)
does particularly well, with almost no artifacts.

I wouldn't be surprised if both used the same resampling code as sox. This
is based on what used to be called the "Secret Rabbit Code" and can do
arbitrary rate conversions quite well. However I tend to use sox for rate
conversions, filtering, etc. Just use Audacity for dealing with clicks.

FWIW I did do some sample-by-sample comparisons on the results of using
Audacity to do this on 96k/24 files. And confirmed that the results were
identical except for the 'repaired' sections.

I'd probably also use 88.2k if the end-aim was Audio CD. But since I'm
leaving the results as 96k/24 (in flac) this isn't a concern for me.
However I'd agree that 2:1 ratio conversions are relatively simple to do
well. What may be unknown, though, is how a given ADC operates when
outputting different rates. Some may run at a high *fixed* rate and do
their own internal downsampling. In such cases you may be better off using
a sample rate for the capture that is a simple scale factor down from that
internal rate. Devil in the details.

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html