jitter reduction?

Here is a paper on the subject:
Eric Benjamin and Benjamin Gannon, "Theoretical and Audible Effects of Jitter on Digital Audio Quality", 105th AES Convention, San Francisco, September 1998. This paper concluded that the threshold of audibility of jitter on normal music signals is around 20nS.

Since the average cheap CD player is around 1 ns then in theory one should be fairly safe even with a portable player for next to nothing. In practice this is not good enough for most audiophiles however, and jitter has become a hotly contested subject. Many people claim to hear jitter and can tell you what it sounds like.

Dave's assertion that a "DSP preamp will mask or strip the music of its essence while adding noise" so as to make jitter inaudible in two channel music does not seem to be supported by the authors (they tested a lot of players and interfaces). To me this sounds like an "urban legend" - there are a lot of DSP's amps out in the market today in HT setups and I don't believe there is such a universal problem with them all. Anyway, I won't get drawn into a discussion on opinions to which there is no conclusion, instead I leave you to make up your own mind and suggest you get the paper ($20 at AES website), I believe you will find it quite readable and interesting.

I can't send you the paper as you need to buy it from the AES, however, I can extract the conclusion;

The effect of clock jitter in the digital interface was studied extensively. Measurements of the jitter spectrum of numerous digital audio sources, primarily DVD players, were conducted. A wide range of performance was found. The jitter spectrum of a typical source can be characterized as a white noise floor with one or many sinusoidal jitter components with a magnitude in the range of 10 ps to 10 ns rms. The effect of jitter induced in the interface was studied and found not to be a significant factor for short interconnection runs likely to found in a domestic environment. Several DACs and their DIRs were measured in order to characterize the sensitivity to distortion induced by jitter. These results were compared to each other and to results derived from simulations. Most DACs were found to be similar to each other and to the simulation in terms of susceptibility to jitter induced distortion. That distortion is approximately -107+201og(F)+201og(J) dBr for sine wave signals at F kHz with J ns rms of clock jitter. Up-Down threshold and AB comparison listening tests were conducted to determine the threshold of audibility for jitter-induced distortion. The threshold of audibility for pure tones was found to be about 10 ns rms at 20 kHz and higher at lower frequencies. For nearly all program material no audible degradation was heard for any amount of jitter added below the level at which the DIR lost lock. Certain program material was found in which an audible degradation due to jitter was heard. The threshold of audibility for these programs was generally found to be in the range of 30 ns rms to 300 ns rms for sinusoidal jitter. Finally, the audible degradation was found to correspond to measurable
changes in the spectrum of the program material.

The influence of jitter in causing audible distortion was found to be less than anticipated by the authors, and less than that predicted by both the technical and consumer audio press. Jitter induced by the digital audio interface was not found to be an audible problem for any of the program material auditioned.

It should not be assumed that jitter-induced distortion is a non-issue. Distortion induced by jitter is a real phenomenon and work to reduce its effects should continue. Although the threshold of audibility was found to be relatively high in the authors' experiments, the effect of all distortions in the audio chain is cumulative and it is reasonable to reduce them to the lowest practical levels. Manufacturers of DACs may find the methodology for evaluating jitter susceptibility presented in this paper useful in characterizing and presenting meaningful jitter specifications for their products.

Acknowledgments
The authors would like to express their appreciation to Edmund Chu, who designed and built the wide bandwidth AB comparator system used in the listening tests. We would also like to thank our many colleagues at Dolby Laboratories who participated in long and difficult listening tests.

As Dave and Steve have pointed out it is indeed a complex issue. Dave claiming to be able to hear jitter quite easily because of his highly resolving non-DSP system. The extensive lab listening tests done by the authors was only a miniscule percentage of the population - so it certainly leaves a "crack" in the door open to the "golden eared" with highly resolving systems. The authors of the paper (Dolby Labs) say they were surprised at how high the level of jitter needs to be in order to be above the threshold of audibility, however, they sensibly recommend maintaining jitter as low as possible, "it is reasonable to reduce them [jitter] to the lowest practical levels."

On reducing jitter as being a good thing everyone is in 100% agreement!

Chris,

The authors found experimentaly through listening tests that pure tones made it easier to detect jitter. They looked at all frequencies not just 20 Khz. The paper is a mix of mathematical modelling and lab measurments and lab listeting tests. Jitter creates sideband intermodulation distortion - new frequencies appear as the jitter frequencies modulate the musical or primary signal. According to the paper, the best way to hear distortion is to get the sideband distortion in the ears sensitive 1 to 4 Khz range whilst keeping the music outside this range so there is as little masking as possible.

I'd strongly recommend to download and read it if you are interested - refreshingly absent of any formulas and complex mathematical jargon that you often see in AES papers. They used headphones in listening tests so loud speaker distortion was not an issue (most loud speaker distortion would probably make it even harder to hear jitter)

The important thing to walk away with is that it still recommends that you should always keep jitter as low as possible as distortion is cumulative...what might not be audible distortion from one root cause can accumulate with other non-linearities to become audible.

I don't doubt that a ribbon would likely make jitter distortion all the more audible. I am just surprised that a ribbon would be regarded as a reference source for listening for jitter distortion. Dome tweeters in general have much lower distortion. Given distortion is cumulative would it not be better to use a dome tweeter?