[time-nuts] Oh dear

Poul-Henning Kamp phk at phk.freebsd.dk
Mon May 7 17:56:14 UTC 2012


In message <D2251F0F290D4B1AB54E1A4DBA34500E at vectron.com>, "Bob Camp" writes:

>If you extend the bandwidth down low enough (as in low audio) the jitter
>goes up quite a bit. In the case of audio, jitter at low frequencies just
>might be something to worry about.

Not with the kind of physical laws I live in.

At low audio frequencies, say 100 Hz, you have at least 441 samples
per period of audio, and the Y-difference from one sample to the
next is so small, that no amount of jitter will have sonic impact.

At a 20 kHz frequency however, you have sign reversal from sample
to sample and moving a sample in X has very high impact on the
energy of that and the surrounding samples.

This is exactly why we use oversampling in the first place:  You
get more gentle slopes from sample to sample which means that
the jitters effect is attenuated in the result.

The place where this audio-jitter-homoepathy comes from, is the
first generation of Philips CD players, CD-100 etc, which had
"jitter" come up from the poor mechanics, because there were
insufficient buffering before the de-interleaver.


-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.



More information about the time-nuts mailing list