Jitter Correlation to Audibility

[robertsong]

Then using your criteria, putting photographs of yourself in your freezer also improves the sound of your audio system as well as any other audio system you might be listening to. Have you any photographs of yourself in your freezer? If not, why not?

se

 
You are definitely going to have explain that analogy to me. Sorry for being so thick.

 

[Steve Eddy]

You are definitely going to have explain that analogy to me. Sorry for being so thick.

Because when you lack any controls, and just go by whatever people purport to "hear," then anything can be shown to make a difference, including photographs of yourself in your freezer (a Peter Belt tweak that some have tried and use, such as Carol Clark of Positive Feedback).

se

 

[esldude]

   
 
Have you not tried the test for yourself? Why not?
 
Step 1: Hook up DAC with an outboard USB/Spdif converter using a spdif cable.
Step 2: Hook up a different outboard USB/Spdif converter using the same usb and spdif cable, and observe the differences.
 
 
This is indeed a controlled test.
 
 
There has got to be literally thousands of people who will testify in favor of audible jitter. For a good laugh, go and post this same proof request on Computer Audiophile, Audio Circle, or Audiogon, and count the number of skeptics.  There is absolutely no need to get into highly technical arguments. And the only person here who has demonstrated any kind of technical knack on this particular subject is Digitalchkn.

So did you control for output level? Were the different outboard DAC's switched blind or sighted?  And if they were, how do we know automatically the difference is jitter? 

 

[SilverEars]

http://www.ifi-audio.com/en/iPurifier.html
 
Here is the discussion on it
http://www.head-fi.org/t/683778/350cad-herus-asynchronous-usb-audio-2-0-24bit-352-8ks-s-dxd-and-dsd64-128-2-4vrms-headphone-dac/135

 

[bigshot]

More viscous liquid derived from petroleum emitted by serpents.

 

[Steve Eddy]

More viscous liquid derived from petroleum emitted by serpents.

Well, the device does have some L/C filtering on the +5V line and what looks to be a common-mode choke on the data line, for whatever that's worth.

se

 

[SilverEars]

What do you guys think about these videos from ESS technology?
 

 

 

[castleofargh]

Can I add oil to the fire by doing the noob(again)?

 
I find the hypothesis: "any changes in sound from 2 sources giving digital signal can only come from jitter", to really need to be confirmed first. must I understand that you would call jitter any kind of change that would occur to the digital signal?

 
Could a difference in source impedance lead to some matter of jitter? Say that voltage ends up going higher into dac C because of impedance difference, couldn't it change the moment when switches from 0 to 1 or 1 to 0 are triggered? (pure guessing here so let me know when I've reached bullshiiiit mountain)

 
Jitter can only lead to the sample being moved on the time-line while quantization error can only move it in amplitude, is that right? But both would still affect the sinusoid in both axis as it would move the possible path going from that sample to the next. in any case it would mostly affect high frequencies right? Because for lower freqs the duration of lag becomes less and less significant.

 

 
What max change could it do in amplitude on, say worst case scenario, 20khz and 300ns jitter that I've seen thrown into a post here ? Again I guess only the highest frequencies could be really affected as the others would have a lot more points to create the signal wave so the impact of one error should be reduced.
For pitch changes I guess I can manage alone, let's take the 300ns jitter. it makes 0.0000003s.
Worst case scenario again 20khz = 0.00005s . so max change would be + or minus 0.0000503s or 0.0000497 so between 19880hz and 20120hz .
is that right?

 
for me it's 5am so I hope the jury will take it into account before delivering its verdict.

 

[esldude]

  Can I add oil to the fire by doing the noob(again)?

 
I find the hypothesis: "any changes in sound from 2 sources giving digital signal can only come from jitter", to really need to be confirmed first. must I understand that you would call jitter any kind of change that would occur to the digital signal?

 
Could a difference in source impedance lead to some matter of jitter? Say that voltage ends up going higher into dac C because of impedance difference, couldn't it change the moment when switches from 0 to 1 or 1 to 0 are triggered? (pure guessing here so let me know when I've reached bullshiiiit mountain)

 
Jitter can only lead to the sample being moved on the time-line while quantization error can only move it in amplitude, is that right? But both would still affect the sinusoid in both axis as it would move the possible path going from that sample to the next. in any case it would mostly affect high frequencies right? Because for lower freqs the duration of lag becomes less and less significant.

 

 
What max change could it do in amplitude on, say worst case scenario, 20khz and 300ns jitter that I've seen thrown into a post here ? Again I guess only the highest frequencies could be really affected as the others would have a lot more points to create the signal wave so the impact of one error should be reduced.
For pitch changes I guess I can manage alone, let's take the 300ns jitter. it makes 0.0000003s.
Worst case scenario again 20khz = 0.00005s . so max change would be + or minus 0.0000503s or 0.0000497 so between 19880hz and 20120hz .
is that right?

 
for me it's 5am so I hope the jury will take it into account before delivering its verdict.

Probably some can do a better job than I.  But I will take a shot at it.  Jitter is as you say a change in the timing of samples.  Its result is noise not in the original signal.  That can take many forms.  As you surmise, timing differences cause greater phase shifts at higher frequencies, and therefore effect those more if it is random timing jitter.  But some jitter is concentrated at other frequencies though a given amount effects highs more than lows.
 
One way to look at it is how much jitter created noise would be enough to change the resulting value by the least significant bit.  For 16 bit that is about 20 picoseconds at 20 khz.  The amount needed to change a lowest bit value would double for each octave lower you go.  Now that does not mean you will hear that change.  Only that such a level of jitter could cause enough noise to change the result by the smallest amount.  Blind listening tests show it takes a far, far higher levels to actually be audible.  Even more so with music vs test tones. 

 

[bigshot]

44.1kHz isn't faster than humans can perceive timing errors?

 

[stv014]

One way to look at it is how much jitter created noise would be enough to change the resulting value by the least significant bit.  For 16 bit that is about 20 picoseconds at 20 khz.

 
At 20 kHz, 20 ps of sinusoid jitter, even assuming that it is 20 ps RMS (worst case, peak to peak would be 2.828 times better), would create sidebands at -115 dBr level (so it is -112 dB total RMS level). That is quite a bit less than 1 LSB at 16-bit resolution. Did you possibly mean 20-bit ?

 

[jcx]

 44.1kHz isn't faster than humans can perceive timing errors?

yes but it depends on your understanding of the definition, understanding of reconstruction filtering
 
the naïve idea is that because the 1/44.1k = ~22 us sample interval is larger than the lowest demonstrated interaural time delay that can be shown in DBT there isn't enough "time resolution"
 
however that ignores the reality - its easy to show that with proper reconstruction filtering that Nyquist really does work - you can represent a bandlimited signal accurately by discrete sampling at > 2x the higher frequency content
 
Nyquist actually "working" includes the phase, differential phase/relative time delay between 2 digitized samples being accurately preserved
 
 
there is a limit caused by the noise introduced by quantization - but you can encode, recover tens of ns differences at moderate amplitude with 16/44: http://www.diyaudio.com/forums/analogue-source/245555-temporal-resolution-6.html#post3697702

 

[castleofargh]

One way to look at it is how much jitter created noise would be enough to change the resulting value by the least significant bit.

I like that. find how much the timing difference would actually impact on the amplitude instead of keeping it in the time domain, so it can be summed up in db/bit value. it does indeed put things into perspective, as I realise that changes will be of very small value compared to the track playing.
so with an estimate value of jitter we could pretty much see it as the dac losing a few bits, and estimate the damadges from a quantization error point of view right?(I'm going wild again ^_^)
 
but then even if some jitters are more noticeable than others, if we know the range of values of that jitter, couldn't we apply some kind of dithering to generate noise at slightly lower bit levels and practically attenuate those "maybe noticeable" jitter noises? (indeed as long as it's low enough so that we don't notice the dithering itself)

 

[ab initio]

  I like that. find how much the timing difference would actually impact on the amplitude instead of keeping it in the time domain, so it can be summed up in db/bit value. it does indeed put things into perspective, as I realise that changes will be of very small value compared to the track playing.
so with an estimate value of jitter we could pretty much see it as the dac losing a few bits, and estimate the damadges from a quantization error point of view right?(I'm going wild again ^_^)
 
but then even if some jitters are more noticeable than others, if we know the range of values of that jitter, couldn't we apply some kind of dithering to generate noise at slightly lower bit levels and practically attenuate those "maybe noticeable" jitter noises? (indeed as long as it's low enough so that we don't notice the dithering itself)

Jitter is even less audible than that! You only have jitter artifacts when a signal is present and the dB level you calculate is relative to the signal level. Even though jitter is worst at high frequencies, the high frequencies only have small amplitudes in music.
 
Cheers

 

[jcx]

different DAC internal operation also changes jitter products - the usual calculation is for "NRZ" (non return to zero - between samples) Zero-Order Hold types - where the output is directly proportional to the clock interval width
 
but there are "RZ" equivalents Capacitor ladder/array DAC dispensing a charge that doesn't depend on time errors between clock ticks