Jitter Test: Can you hear it? (downloadable samples) - Page 2

Jitter only cause error when it exceed the tolerance of hardware. Jitter is not error.

I think you misunderstand. It is impossible to design a perfectly accurate clock, ergo errors in timing between the sampling points (jitter) will always exist. What causes these errors (jitter) to become audible artifacts is when they cease to be insignificant to our ears. The point at which these errors (or jitter) becomes audible is still, I believe, hotly debated. Audible or not, the point is, that timing errors (jitter) always exists to some degree and are encoded into the signal when it is first recorded (sampled) by the ADC. The sample data is recorded into an audio file, so this audio file does contain jitter (timing errors).

G

So any of you golden eared folks been able to pass the test so to speak?

Or are you just arguing over it?

I (tin ears) tried it, no chance, but I did mathematically work out which was which for the second set of samples

That explains a lot. I didn't know ALAC was also in a m4a container. I don't have the ALAC component, so it won't play. I'm not planning to install it either

If you can hear any difference in the first 4 files, then you have a resolving system.

I picked them correctly through the noise, but switched the 0nsec and 10nsec tracks.

Could not pick the second set of 5 correctly, although one of my customers nailed it. Some of my customers have megabuck systems. $5K speakers? Try $50K.

Steve N.

Steve, with respect there were 4 samples and 4 slots and you got 2/4 correct. This is not picking them correctly as I see it. By random guessing alone you get 1/4

With 4 tracks you have a 25% chance of getting 2/4 correct, this is not statistically very powerful.

Also you are on record saying that you can detect 2ps of jitter so how come you thought the 0ns jitter track had 10ns jitter and vice versa.

The 2nd set had a track with 100ns jitter and one with 0ns jitter neither of which you got in the correct slot.

Could it be that jitter really is not that important at such levels ?

As for your customer nailing it well in fact

he listed 2---4----3---5---1 as low to high and the order was 2--(5/3)--1--4 so he got 2/5.
actual-- 0--100---10--10--30

Could you ask your customer to do some blind tests as that would be much more credible.

I "nailed" it, but I cheated by analyzing the waveforms mathematically.

I am running a set of DBTs in FooBar between the 0ns and 100ns files. I found a tiny section of the Joni Mitchell track which seemed slightly different in the two files round about 15 seconds in and about five seconds after the vocals start, I had to go backwards and forwards again and again and again. I am currently at 11/13 (p = 0.033) but I had to go home so have not reached a decent level (20 tests) yet, I will get back to it tomorrow. Frankly I am surprised at this result my system must be more revealing than I thought or I could just be a lucky guesser

But even if I do get 18/20 it is 100ns a figure that is so much higher than even my ten year old DAC and 20 year old CD players that it is laughable to worry about.

now that I've switched from the STX to the ST(which has a 0.5ppm clock), I can definitely say that this test is m00t!

it's only adding -dubious- jitter on top of the existing one...and it's a constant jitter that's been injected in these files.

like these FLASH videos that try to show you what DLP rainbows look like....a very low jitter gives a very accurate stereo picture, it might even be disturbing because your brain is not used to this kind of accuracy(not mine anyway).

I fully agree w/ what burson says here : Burson Clock

the HydrogenAudio skeptical integrist approach doesn't always work

@LeePerry, weren't you running from thread to thread with this test, telling everybody to try it and because of this test jitter is not audible?

.