Very interesting reading, I'm looking forward to parts 5 and 6. Thanks for the heads-up DUP!

My engineer's interest in the science behind jitter is raising lots of questions in my mind about the specifc relationships between audio jitter (quantity and type/'quality') and sound quality. For instance:
- What levels/types produce which audible artefacts?
- Are certain types of jitter more audible/disturbing than others?
- Objective and subjective comparisons (and their correlation!) of different jitter-reduction strategies (e.g. PLL, I2S etc.) on resulting analogue waveform/sound.

Judging by the titles of the papers there seems to have been a fair bit of discussion of this at recent AES conventions but since I'm not a member I don't have free access to the reprints. I don't mind paying to download a couple of the really good ones but I'm not going to pay to download them all!

Can anybody point me to some really good pieces on the above topics? I am looking more for research papers with some degree of academic rigour, not just narrative/opinion pieces and certianly not vendor hyperbole.

Quote:
Can anybody point me to some really good pieces on the above topics? I am looking more for research papers with some degree of academic rigour, not just narrative/opinion pieces and certianly not vendor hyperbole.

These articles might be helpful:
Jitter, Bits, & Sound Quality by John Atkinson

Jitter & The Digital Interface by R

This may sound like I'm being difficult but I really want to know.
What is the difference between jitter and dither?
I know one is done on purpose and the other is a negative fact of nature. I suppose one is controlled and the other is not.

Jitter is a timing error. The right sample at the wrong time, is really the wrong sample. Imagine the example below represents a digitally encoded waveform, with time going from left to right:

/\_/\_/\_/\

The next example is the same piece of music, but, the timing is off:

/\______/\/\__/\

Clearly, this would not sound the same.

Here's my layman's understanding of dither. Dither is random noise that tricks a DAC into reading the encoded information with greater accuracy. The Least Significant Bit can sometimes get truncated -- say the LSB falls between 2 sample points. A DAC will read the next level up and ignore that bit. By adding dither/noise, the DAC is fooled into reading the next level down and includes the bit that fell between samples, resulting in better linearity. It's counterintuitive, but, by sacrificing the signal to noise ratio, you actually resolve more information.

My understanding is that dither makes quantization errors random; that is, by intentionally introducing noise we do not hear the errors. For example, on a long quiet reverb tail one can hear the level jump up and down as the level rides on the least significant bit. By adding dither (noise) we no longer have the jumps.

The ear is less sensitive to random noise. To make it even less sensitive the noise can be shaped so that it occurs less in the frequencies where the ear is most sensitive.

Its a moot point if one uses a DAC that takes over the transport clock. That it seems is the ultimate solution to jitter.

Jitter is a timing/frequency issue?
Dither is a amplitude/level thing?

It seems to me that at the relatively low sample rate of 44KHz that jitter would have to be really really poor to miss a sample.

Adding noise to correct for level ambiguties sounds like turning up your car radio because the engine was making funny noises.

Either I am missunderstanding the discriptions in the posts or they are too oversimplified.

I'm sorry you found my distillations unsatisfactory. Here's something that might suit your needs:

Dither Explained HTML

or

Dither Explained PDF

Edit: I just found the original source:

http://www.cadenzarecording.com/papers.html

For jitter, the Stereophile links in the previous post by Stephen should cover a lot.

cyclebrain,

Here is one of the best explanations of jitter and its causes that I have come across.

The animated GIF is a great example of a picture being worth a thousand words; I suspect this is the effect Jeff was trying to achieve from within the confines of HTML

Stephen,

Many thanks for these references. Coincidentally I reread the R

cyclebrain,

A couple more links that might be of interest. Jitter and dither are a couple of good grounders by mastering guru Bob Katz.

Also, found a couple of interesting articles related to my original query above. Passing on for anyone who is interested.
Specifying the Jitter Performance of Audio Components
Detection Threshold for Distortions Due to Jitter on Digital Audio
Jitter Simulation in High Resolution Digital Audio
High Resolution Audio DACs - Final Report

Finally, if anybody happens to have a copy of the Benjamin and Gannon paper "Theoretical and Audible Effects of Jitter on Digital Audio Quality" (AES Preprint #4826) please shoot me a note.