OK, looks like Arny, wisely doesn't want to answer what jitter sounds like. The answer as I mentioned was that you can't answer that question.
Earlier someone post my graph with two simple sidebands showing (thanks by the way
). That is for an artificial scenario of one and only one sinusoidal jitter. Here it is again:
Now let's look at the spectrum of a real device's jitter as measured over its HDMI input by Paul Miller of HiFi News:
Doesn't look so pretty now, does it? Instead of just one spike, we now have countless others. Do you think the audible distortion is the same between the two? Of course not. All of those sidebands *combine* their energy and that is what you will hear as distortion, not individual spikes in isolation.
Would another device have the same set of jitter components? Of course not. Why do we have so many? Because there are many sources for jitter and some are square waves and hence, have a ton of harmonics. Think of a microprocessor inside your player waking up once every 1 millisecond to do some housekeeping chore like keeping time. If the CPU is idle and then suddenly starts doing that work, and it is that activity that induces jitter into the clock, the jitter source is squarewave. It is a pulse that goes up rapidly and stays high as long as there is CPU activity. Once that "time out" is done, the CPU goes idle. Looked as a spectrum, it is the squarewave and such a waveform has infinite odd harmonics. So a 1000 Hz (1/1 msec) becomes 1000, 3000, 5,000, etc. Now add to that the front panel display having a refresh, various video devices inside doing their thing, the disc mechanism, etc. They all contribute in some ways to such jitter. Of course, some get filtered and many are attenuated but how much and how well, is equipment dependent.
Take another kind of jitter: random. Here the jitter source is all over the map and doesn't have clean spectrum. Think of a CPU constantly running now. Non-intuitively, this kind of jitter is the least harmful since all it does is create random sidebands and random sidebands is another name for your noise floor.
Then there is the matter of psychoacoustics. That science determines how much of jitter is audible. Arny keeps talking about jitter in analog devices. I have debated that with him and he finally conceded the argument there. In a nutshell, in analog world "jitter" has very low frequency and as such, it heavily gets masked by the music. This is why despite having a lot more of it, it is not objectionable. Arny talks about having tested jitter. Well, because of that misunderstanding, he only simulated low frequency jitter. Here is an example archive from his site: http://home.comcast.net/~arnyk/pcabx/jitter_servo/index.htm
. You see him showing *5* Hz jitter. 5 Hz! Why would we want to worry about 5 Hz jitter? That distortion would be in the shadow of the music signal that generates it and as such, inaudible except in huge amounts. Listing to huge amount of 5 Hz jitter does not tell you what "jitter sounds like" in the levels and frequencies that occur in digital systems. Yes, digital systems also have 5, 60 Hz, etc. that that Arny simulated but for the most part, we don't worry about them. It is the higher frequency ones that get past the skirts of auditory masking and make them potentially audible.
So it is all about spectrum. If you don't know that, then you don't know what jitter sounds like. The fixation on the single number jitter value is very poorly placed. We use it to talk about the concept but the reality is that it is the spectrum that determines the level of badness, not that one number.