[linux-audio-dev] Sample Rate Converter Comparison

Denis Sbragion d.sbragion at infotecna.it
Tue Jan 23 17:35:12 UTC 2007


Hello Erik,

On Mon, January 22, 2007 22:53, Erik de Castro Lopo wrote:
> SecretRabbitCode was recently included in a test of a number of
> commercially available sample rate converters and while it wasn't
> the best, it certainly didn't disgrace itself either.

I once tried to further improve your sinc code, using longer and better
filters optimized using the Dolph-Chebyshev window and a different balance
between bandwidth and stopband attenuation, of course at the expense of
performances. It turned out that I couldn't improve the results above a given
threshold because of a sort of (really low level) jitter in the timebase
calculation. With integer resampling factors, where the jitter dind't appear
because of the way the timebase is computed, the results where on pair with
CoolEdit/Audition, which is among the best in that test.

It was long time ago, so I don't remeber exactly what I did, but this
information might be useful in improving your code. Removing this timebase
calculation problem, if it can be removed at all of course, probably will make
your code close to perfection, making the final performace just a factor of
the quality of the filters used. Regarding this, you might think about an
option to load the filters from file... :)

Of course this is just academic, because I really doubt that the tiny
artifacts introduced by SRC are audible at all, even in its current
implementation. Just a matter of winning the race. :)

Bye,

-- 
	Denis Sbragion
	InfoTecna
	Tel: +39 0362 805396, Fax: +39 0362 805404
	URL: http://www.infotecna.it




More information about the Linux-audio-dev mailing list