I'm not sure that all serious people would agree with you about audible thresholds. But I do think over the past 10 years things have improved a lot. DAC chips are better. Clock cleaner circuits are now available in ICs, etc. My Cranesong HEDD and Lynx 2 are a good 10 years old, but both are still in production. Sy, they sound different. I don't have special ears. You could most likely hear it yourself if you have normal hearing, at least if given a little training and practice.
Anyway, I'm not sure what numerical values you mean when you say "orders of magnitude." I don't have the AES papers, but at this website: Jitter in Digital Audio Data Streams Article By Steve Nugent Of Empirical Audio They summarize some published results as follows:
Studies by the AES (analysis, not human testing) conclude that these are the thresholds of audibility:
[1] 120psec P-P jitter audibility threshold for 16-bit DAC and 8psec P-P jitter audibility threshold for 20-bit DAC
[2] 20psec P-P of data-correlated jitter audibility threshold at certain frequencies and "A simple model of jitter error audibility has shown that white jitter noise of up to 180psec P-P can be tolerated in a DAC, but that even lower levels of sinusoidal jitter may be audible"
Steve Dove once did some consulting for Jack Tramiel on a CD player. He said it was so lousy that it had audible time base errors, a sort of wow and flutter.
This sort of thing reminds me of an AES LA Chapter meeting in which some of the very-earliest barely-functional internet streaming audio was presented. It sounded ludicrously terrible, really not even amenable to description in any conventional audio measurement terms. I couldn't stand it anymore ("Mandrake, I cannot sit back and allow...") when one of the presenters said But this will get better---this is only 8 bit encoding! I said Buddy decent 8 bit dithered digital audio is super-HiFi compared to this!
However we all knew it would get better. At the time it seemed churlish to complain too much, rather like criticizing the bear riding the bicycle for being less than elegant and graceful.
Who said that they are indistinguishable? The reason for doing a controlled listening experiment in this case would be to test if a specific listener is really able to differentiate the two loudspeakers. (For SY without peeking) .
I encourage everybody who believes that it must be an easy task to try it.
Sure. Sometimes people prefer their favorite distortion. I like varying amounts of it on my electric guitar. I don't claim it is Hi-Fi, though.
Well, I just got it all dialed in (maybe) and have not tested that thought yet. But, I certainly will get there soon. Cover up is not something I am into. Uncovering is something I am interested in. So far, the CD's DO sound better. Most noticeable, with the last few speaker systems, is the dynamic range is greatly improved. The sense of annoying compression compared to live music ---The lack of dynamics is mostly gone, now. Was it the amps or speakers? or both? Later, when I change amps, then i can answer that, too. But, room interaction has greatly been reduced and that is a good thing as well. Over-all, a step up towards more real sounding.
Ever better and more realistic sounding. Jam from N.Pass Labs is coming over soon today. We'll listen together. He is also bringing his new Headphone amp.
THx-RNMarsh
Last edited:
Markw4, it should be obvious that an observer can believe in your description of the perceivable difference (due to audible differences) but does not to have to.
Because it could be that you only think that an audible difference exists. That´s the reason for doing controlled listening tests to ensure that most likely the independent variable is the reason for the difference.
Wrt jitter the AES analysis afair relied on theoretical considerations regarding to different jitter mechanism and hearing curves without considering masking effects.
Real listening tests had up to now results above 1 - 3 ns (deterministic jitter with worst case sinoidal components).
In reality things are a bit more complicated but more refined tests are missing...
My recollection is that with TTs it's more common to refer to things like wow and flutter, but I guess you could call it jitter.
Regarding CDs, when CDs first came out there was a rush to digitize many analog recording to sell everybody a now copy of their favorite music. The ADCs in use at the time were not up to today's quality. They sound bad to many people, and jitter is only one of the problems.
Moving along in time, any music recorded on the early Pro Tools systems were poorly digitized by today's standards. They sound awful to me, but maybe you feel differently. Again, jitter was only one problem.
In more recent years, dual cascaded PLLs became more common to clean up clocks, but the PLLs were not easy for some designers to do well. Now we have ICs that do it very nicely.
What does it sound like? It can sound a lot of different ways depending on the particular jitter spectral content. But the things people tend to notice most have to do with stereo imaging and harder to define things like muddiness, time smearing (kind of like some FIR filters on percussive sounds where you have pre-distortion in the time domain), or sometimes there is some increase of perceived clarity. The latter one can sound like some kinds of harmonic distortion, and in fact harmonic distortion generators are sometimes intentionally used to increase perceived clarity of muffled sounding recordings.
The easiest one to demonstrate is imaging. If you generate sounds in a computer, then pan them in a mix, you can try to close your eyes and point to the spot in front of you where you think the instrument is in the sound field. With very low jitter and properly set up speakers (in the near field) you can point to the spot between the speakers pretty well. With more jitter the spot get spread out in perceived width, until finally it sounds like it is coming out of two speakers and not from somewhere in between them.
Last edited:
Glad that this was brought up.
Perhaps the pink elephant in the room is going to be "imaging" along with "soundstage", "ambience" and the other terms that are used to describe how the sound is perceived.
It's the one thing that has nil correlating whatsoever to any parameterized measurement(s).
Thanks, I don't listen to a lot of horns but I suppose there might be some that don't have obvious non-minimum phase artifacts that also can not be equalized except by some kind of DSP exercise (?).
For what it's worth, tell Jam I said hello. He hates me now so he might not take it so kindly.
se
Cites, evidence, or maybe you are familiar with all measurement techniques ever conceived? You did say nil = 0 right?
Well, I call it jitter as it was the result of bad PLLs controlling direct drive turntable motors. Lower freq. jitter, I grant you!
I've never heard a system with sufficiently bad jitter that the stereo image was smeared because of that... And I've listened to many systems over the years, low and high end, low cost and megabucks! ........maybe I've just been lucky!
Incomplete, but not incorrect. Now if one could actually compile a complete list, that would be a truly amazing accomplishment. That also points up why assigning numbers indicates direction but won't get you all the way there.
Scott,
I do know what Bob Pease said when developing the LM4560 family. The listening tests actually surprised him. Folks with a bit of high frequency loss due to age were often better than younger folks with better measurements for hearing loss.
You should have heard the AD/DA extension card I developed for my Apple II clone in the middle of the eighties (I don't think you had those in the US, but in the far east they were easy to find). Intelligible sounds could actually be recorded and reproduced, but the way it sounded still gives me the creeps.
I find it makes difference to me. For example, I record a male vocalist playing an acoustic guitar. He is singing some of the same notes the guitar is playing. Around 300Hz, where the ear is sensitive to that, it can get to sounding muddy. If we have EQ and panning available to deal with it, too much of either one doesn't sound right. Same for too much overlap (too little EQ or panning) at that frequency. What I find is that if I make a slight adjustment to panning, then I have to compensate with EQ to find the best sounding mix subject to the constraint of the panning decision. If the panning is smeared, it makes fine tuning the decision much harder. In this case, as little as 0.1 or 0.2 db change in EQ may be audible. Not always, or even usually, though. More likely to be so if using a bus compressor, since small changes in mix balance seem to change the response of the compressor. It's not that the 0.1 db or so has a direct effect I can hear so much as to whether the compressor is causing the guitar to duck under the vocal, or vice versa.
The recording I am thinking of, I am told by a former student and the male vocalist in question, is still being used at Eastman music school in the recording studio, as an exercise for students to try and figure out how to get the particular guitar sound.
- Status
- Not open for further replies.