|
Propeller Head Plaza Technical and scientific discussion of amps, cables and other topics. |
|
|
|
Propeller Head Plaza Technical and scientific discussion of amps, cables and other topics. |
|
In Reply to: Mostly Jitter Related posted by Jon Risch on May 15, 2005 at 21:12:27:
Alright,Lets get some of the thoughts about jitter going here and see if we can figure it all out.
1) Someone posted a link to a stereophile article (below). I think the graphs are extremely misleading on waveform distortion because the actually don't seem to take into account that the signal should be band limited before sampling (so it cannot have discontinuities) and the reconstruction filter will never produce a signal with discontinuities. This is a very important distinction for what comes next.
2) The ideal non causal reconstruction filter (frequency brick wall) is a very non-inuitive mathematical constuct. Jitter measurements with samples of a steady state sine wave make sense for measurement pruposes becuase the sidebands produced give some indication of continuous jitter. Unfortuntely, typical sounds, instruments, pieces of music, etc are not a steady state sine wave. Also, the total jitter on a clock signal will be partially periodic, partially random, and partitally signal related. This is an exteremely complex situation.
3) We really do not know that much about how we hear. We know the transducers work in the freqeuncy domain and produce a time modulated signal (the rate of firing of each frequency sensor) that represents frequency magnitude and possibly phase (and phase deduction may also be due to different measurements in two ears). Beyond that we have no idea how we isolate individual sounds in a complex signal, but it must be one hell of a pattern recognition engine in there.
4) There is a lot of talk that jitter is FM, but it is FM of the digital clock. This does not tell us the resulting behaviour of the analog signal after the reconstruction filter.
5) Although I haven't done the math, I suspect jitter does a lot more than add spurious frequencies to a signal. It can also change the magnitude and phase of the frequencies that are supposed to be reproduced and the relationship of associated harmonics and the harmonic structure pattern is how we recognize individual sounds in a complex signal.
When you consider that a set of samples _uniquely_ identifies a band limited continuous signal from a given reconstruction filter, changing anything about the samples (like position) identifies a new signal and its relationship to the original is not simple thing (not just extra fequencies or noise) . With enough distortion on the time axis you could possibly end up with samples that violate the band limit of the signal based on the sampling period.
Jitter and hearing are both extraordinarily complex things and most of the articles I've seen don't really address the full complexity of the issue.
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- HANG ON EVERYONE! Lets try again without so much bickering - macaque 08:57:39 05/20/05 (6)
- Re: HANG ON EVERYONE! Lets try again without so much bickering - Tom Dawson 14:43:54 05/20/05 (0)
- Re: HANG ON EVERYONE! Lets try again without so much bickering - Dan Banquer 09:48:30 05/20/05 (2)
- Vagarities of human hearing..... - unclestu52 11:33:27 05/20/05 (0)
- Re: HANG ON EVERYONE! Lets try again without so much bickering - macaque 10:48:08 05/20/05 (1)
- Thanks for getting us back on track - Commuteman 09:44:44 05/20/05 (1)
- Re: Thanks for getting us back on track - macaque 16:03:26 05/21/05 (0)
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
