[tschanrm]

Lately I've been doing some thinking about jitter, and the effects it has on the audible signal of sound. From doing some reading, I found a website which has test signals with added jitter (also called fm modulation on the website). Read this page, then download the test samples. There is a link on the bottom of the linked page for the samples.

http://www.pcavtech.com/techtalk/JIT/index.htm

All samples were tested with the headphone out from an Echo Layla 24/96, using my modded Shure E2c canal phones (no waxguard and sony soft tips) and my Brother's Bose Quite Comfort Noise Cancelling Headphones (these were the only headphones I was able to get my hands on). I feel I have a pretty good trained ear, and have done ab/x testing in the past on Hydrogenaudio with mp3 test files. While real-world jitter is more complex than these samples, it is still good fundamental exercise in identifying just what jitter "should" sound like.

I did some ab/x testing from the samples provided, and found it was very hard to tell the difference between the worst samples (-20db & -40db jitter tone samples) and the reference sample. With all other jitter samples (-60db & -80db jitter samples), it was impossible for me to get statistically significant results; in other words I could not tell the difference between the sample with jitter and the reference sample. My point of this thread is to find out if anyone else can detect the jitter in the samples. Most decent quality equipment have s/pdif jitter similar to or lower than the -80db sample, and I could not ab/x this sample at all (I tried many times on seperate days).

My equipment I tested on may not be the best, but I think it is at least mid-fi. From this test, using mid-fi equipment, my opinion is that for music playback jitter is much overhyped, and investing money in a better DAC or headphone amp is a better alternative than investing money into jitter reduction units. Now, for music recording, I feel jitter is much more important, as you need the most accurate signal possible for editing purposes (which is why I feel word clock is important for recording).

Please listen to the audio samples, but please compare them using the ab/x software only (or using the foobar ab/x plugin). I like ab/x because it is one of the best ways to remove the psychological effects of comparing audio samples. A word of advice on testing: be patient, and have fun

 

[thomaspf]

That is an interesting link. I am a bit unsure how this relates to the jitter on a conversion clock in a DAC.

I thought the whole idea was that the analog waveform created by a DAC contains artifical distortions that originate in the clock variations of when the conversion is happening. An identical bit stream will create different analog signals in the presence of different amounts of jitter.

There is a standard J-signal that is designed to make these distortions more easily measurable via standard FFT in which they show up as side bands.

How this relates to different digital streams that have certain bits flipped and whether one should be able to ABX those, I really don't know.

Cheers

Thomas

 

[halcyon]

Thanks for the good link.

Some things to think about:

1) If you use jittery playback sources/devices, how can you ever consider to be able to hear the non-jittery signal?

2) There are various types of jitter: (pseudo)random, data-induced, HF jitter, etc. It would probably be wise to test each of these individually

3) Human hearing has a high tolerance for harmonic distortion (look at the first jitter test on the page). Not all jitter causes harmonic distortion

4) Many of the audible differences caused by huge amounts of jitter are imaging related and much easier to spot on properly imaging loudspeaker (+room) system than on a pair of headphones.

Regardless of all above, I think jitter is indeed often overhyped as a detrimental cause of sound quality, as there can be problems that can cause significantly worse audibility artefacts (signal and audibility wise).

Some of these include typical IV conversion artefacts, bad filtering/noise suppression in the power circuits and the sound of some components (capacitors being an example in extreme cases).

Just my two cents worth.

 

[JaZZ]

Quote:

Originally Posted by halcyon ...If you use jittery playback sources/devices, how can you ever consider to be able to hear the non-jittery signal?

Excellent point! Also I'm generally skeptic towards the idea of simulating jitter within a 44.1-kHz/16-bit data format by means of the very same data format. Logic tells me that it doesn't work.

 

[tschanrm]

Quote:

There is a standard J-signal that is designed to make these distortions more easily measurable via standard FFT in which they show up as side bands.

Thomaspf, are you referring to Julian Dunn's testing methodology? I know stereophile used to use this jitter test signal with a Miller Audio Research Jitter Analyzer. As for the other things you mentioned, I'm not sure what you are trying to get at. Timing error of bits due to the clock create the added harmonics that are present in the test signals (in simplified form), just not to that degree.


Quote:

...If you use jittery playback sources/devices, how can you ever consider to be able to hear the non-jittery signal?

Halcyon,

You should be able to ab/x the worst signal (-20db) with amost any decent source, as long as you have a trained ear. This signal adds 10% jitter (or fm modulation distortion), which should be higher error than any sound card's clock error and higher than any s/pdif signal error as well. The other three test signals, -40, -60, -80, I'm not sure about.

Quote:

Excellent point! Also I'm generally skeptic towards the idea of simulating jitter within a 44.1-kHz/16-bit data format by means of the very same data format. Logic tells me that it doesn't work.

Jazz,

The test samples aren't simulating jitter, it is jitter. This is, in a simplified way, what happens to your digital signal, it is just overly exaggerated in these samples. Jitter is a form of distortion, therefore could be heard at almost any bit depth or sample rate. As an example, look at the picture of the -20db jitter sample, then look at the top picture at this page from stereophile:

http://www.stereophile.com/integrate...91/index8.html

You'll see that they are similar, with the real-life stereophile results having much more complex jitter at lower amplitude levels. Hope this info helps.

 

[bobeau]

Regardless of theory and equipment used, has anyone else who listened to these samples been able to discern a difference?

 

[thomaspf]

I am just making the same point as Jazz. It is unclear to me that by modifying the digital stream to include a signal that might look like jitter on a real DAC and then sending this through a DAC with it's own jitter characteristic you can actually learn a lot.

It is an interesting thought experiment though.

Cheers

Thomas

 

[crazyfrenchman27]

You can hear jitter.

Jitter is very real.

There are sample mp3's out there that demonstrate the difference.

-Matt

 

[thomaspf]

Nobody here doubts that jitter is real. However, embedding jitter in a digital stream is not straightforward.

Assume you are playing back the jitterless original on a DAC with some inherent jitter. You record the analog output from that DAC with an independent ADC that for arguments sake has no jitter at all. If the distortions caused by the jitter are large enough this will change the samples of the track that you just recorded.

Now you take that track and ABX it against the original on the DAC. What would you expect to hear?

What happens if you repeat the this experiment 5 times? Will the jitter get worse?

Cheers

Thomas

 

[clarke68]

Quote:

Originally Posted by tschanrm my opinion is that for music playback jitter is much overhyped, and investing money in a better DAC or headphone amp is a better alternative than investing money into jitter reduction units.

I haven't listened to the test yet (but I will!) but I agree with you, that there are much more effective ways to improve the sound of your system than jitter-reduction units. Since one of the most notorious types of jitter (transport) comes from the s/pdif interface itself, adding a box in the chain that converts the signal from and back into s/pdif is an inherently flawed solution. Much better to invest in DAC that is less suseptible to jitter in the first place.

Quote:

Originally Posted by tschanrm Now, for music recording, I feel jitter is much more important

Again I agree...jitter at the A/D stage is permanent, and can't ever be removed.

 

[bobeau]

Quote:

Originally Posted by thomaspf Now you take that track and ABX it against the original on the DAC. What would you expect to hear? Thomas

Say you're given an out of focus photo. You snap a picture of that photo with an out of focus lense.

Can you tell a difference? Sure, the new photo is obviously worse than the first, even if they're both out of focus. It's a relative not absolute perception.

How is this different?

 

[maarek99]

Quote:

Originally Posted by bobeau Say you're given an out of focus photo. You snap a picture of that photo with an out of focus lense. Can you tell a difference? Sure, the new photo is obviously worse than the first, even if they're both out of focus. It's a relative not absolute perception. How is this different?

So? I mean, if you don't hear a difference even after adding jitter then what's the point in measuring it?

 

[Nandro]

Wouldnt this experiment also depend on the headphones you use to listen? I mean wouldnt 325i's be more able to pick it up than say my SR-60's or regular iPod buds? Hence its not even really relevant till you get equiptment that can discern the jitter?

 

[bobeau]

.

 

[bobeau]

Quote:

Originally Posted by maarek99 So? I mean, if you don't hear a difference even after adding jitter then what's the point in measuring it?

My point is why are folks harping on their DAC's own jitter (whatever that may or may not be) as being a monkey wrench in this test for audibility. This is a relative, not absolute, obvservation.