What does Jitter sound like?

It is like looking through a window that is fogged-up. HF noise mostly riding on the music. The spectrum and amplitude have a large effect on audibility, as does the system. If the system has a lot of sibilance from other sources, then it may be masked by this.

Steve N.
Empirical Audio

Shadorne - the positions of the pits in the CD are a form of timing information that can cause jitter as they are read. How this affects the PLL/buffering in the CD player depends on the design of the CD player. It is evidently important though. Rewriting a disk on CD-R with low jitter is an audible improvement on virtually all CD players I have tried. I mod a LOT of transports and players, so I get to try many different ones. This is good evidence that the pit locations are actually important.

Steve N.
Empirical Audio

Shadorne - you are missing the point. It is not the data in the copy that is different, as explained in the paper you cite. It is the jitter DURING real-time playback that is different. This is the distinction.

Steve N.

Bigamp - Sooner or later the folks that dont understand the jitter thing will hear it or they will eventually upgrade their systems to the point where they hear it. Until then, it is like trying to explain how the earth looks from space.

Jitter is simply inaccuracies in the timing of the bits that make up the data stream. It is like a clock that ticks every second. If the clock ticks at exactly each second time interval, then it is said to have no jitter. If some ticks come at .999 seconds and others at 1.001 seconds, then the overall time will be accurate over many seconds, but there is jitter in the timing. This is how actual real-time systems work. They all have some amount of jitter.

The effect of this jitter on the D/A conversion is to create frequency modulation in the analog output. This means that the point at which the top of the cymbol crash was supposed to occur actually occurs maybe 1 nanosecond later and then the trailing ringing of the cymbol comes maybe 1 nsec earlier. The rate of this change can be anywhere from 10hZ to 100kHz or higher. If there is only one sample that occurs at the wrong time, it will never be heard, but typical jitter signature is usually a constantly changing time error. This is why it is audible. The brain detects these things just like it detects moving objects with eyesight.

The time error has both amplitude and frequency or spectra characterisitics. Each CD player or computer audio device has different amplitude and spectra for its jitter, so they can sound very different from each other, even though the data is always the same. Data errors are very uncommon for both CD players and computer audio. Jitter is the difference that you are hearing, if you hear a difference.

Steve N.
Empirical Audio

Dgarretson - Wow and flutter exhibit the same results as jitter, frequency modulation, so the answer is yes, although some jitter occurs at rates and changes in ways that a turntable could not physically do.

Steve N.

"My point was that there is no jitter added or lost by copying CD-R's many times (provided everything remains in the digital domain). The same is true for digital processing - it adds no jitter. ONLY a jittery clock or jittery signal makes jitter."

I agree with all of this. However, inaccuracies in the pit locations on the CD or CD-R contribute to the PLL clock jitter in the CD player. This has been demonstrated thousands of times. There were even several products that re-wrote CD's to get improved pit placement. They work with every CD player I have tried.

Steve N.

Brownsfan - As I have said in other posts, it is the PRIMARY deficiency that makes Digital Audio fatigueing, and is the single most significant flaw in the S/PDIF and CD format.

Steve N.

Shadorne - how do you think the data coming off the disk is stored in a buffer in the CD player without a PLL?

BTW, I am an EE. I was a design team lead on the Pentium II at Intel Corp. 30 years design experience designing everything from big disk and tape controllers for IBM equipment to massively parallel supercomputers to slot 1 processors.

Steve N.
Empirical Audio

"Isn't the buffer in a CD player designed to store data so that it can be processed/decoded to extract the relevant audio bit stream information so that only these "bits" can be sent to the output of the DAC?"

There is minimal buffering in most CD players, usually inside a controller chip that does the servo control to control the speed of the spindle. There is also a CODEC that decodes the data stream from the Pits, as well as read amplifiers etc..

"Aren't CD data (pits) stored in an error correction format called Solomon-Reed interleave code (not a straight audio bit stream) and therefore the clocking of data from the lazer pick up off the disc is NOT directly related to the clock that clocks data in to the DAC output?"

This is true, the frequencies are not identical, but related.

"Isn't there a separate timing required to control the buffer under-flow or over-flow (not clocked by the same clock that controls the DAC output even if the timing is obviously related as the buffer is ultimately suppling the bit stream)?"

The clock for this is created by PLL, so no overflow or underflow occurs. The Crystal oscillator sets the frequency for the spindle rotation, but there is "slop" in the frequency of the clock due to the PLL. The spindle rotation is not that precise, and the bits coming off the disk have a lot of jitter, so it must track this. There is generally a buffer that attempts to do a second PLL on this and reduce jitter, but it is usually not immune to the jitter coming off the disk.

"Since the buffer "buffers digital bits" then we don't have potential for jitter UNTIL the clocking out in the output of the DAC - or am I missing something again? In which case how does the buffer PLL affect jitter - is it induced noise on the power rails or some other in direct manner?

The secondary buffer output is clocked by a divided-down clock based on the bit-density on the disk (CD versus DVD for instance). In theory it is a fixed clock and should not be subject to the jitter coming off the disk, but unfortunately all of them are. Somehow, the chips that are used for these processes are imperfect, or the power systems are imperfect, so some jitter leaks through. It seems like this problem should have been solved long ago, and it has probably gotten better, but the problem still persists.

All you have to do is rewrite a CD to CD-R and listen to it on ANY CD player and you will hear a difference in a resolving system. Put a mat on the commercial CD and play it and you will hear a difference. Put a treatment coating on the commercial CD and you will hear a difference.

If these CD mechanisms worked ideally, then none of the above treatments would change the sound.

Steve N.
Empirical Audio

Shadorne - the jitter download is coming. It's in the works.

Steve N.

Shadorne - this paper is more meaningless than my jitter samples. The system and music tracks used for this test was a joke. No conclusions can be drawn from this study IMO.

At least with my jitter sample (requested by many on Head-Fi), you can hear the difference between a standard CD and a re-written CD. It will prove that the jitter with your CD player and a commercial CD IS audible if you hear ANY difference.

Steve N.