Jitter vs no-Jitter

[esldude]

Hi Bib

Really interesting. I have just looked at the other sites you published the test on (i deliberately didnt prior to casting my vote) and see that out of 27 votes 19 (70%) thought A2 was the no jitter track. Im no statistician and this wasnt the most scientifically rigorous of tests (bit of fun really), but it would seem to suggest that even though they were wrong, people fairly consistently thought they could hear something different between the two tracks. There was something people fairly consistently associated with track 2 as being better.

What did I hear? Well, in track 2 some of the lower Harpsichord registers there seemed a little more warmth at times, and there was an emphasis on some of the high frequencies.

I suppose if you went on this poll you would conclude people can hear the effect of jitter, but whether they can accurately identify it maybe not ! They seem to like it !


Computer Audio

Track "A1" 4 26.67%
Track "A2" 9 60.00%
I am not sure to hear differences 2 13.33%

Next
song "A1" 0 0%
song "A2" 6 100.00%
I'm not sure to hear differences 0 0%

Well if there were only 2 choices, 19 of 27 is where you cross the 5% likelihood of this being chance. However, having the choice of no difference alters that. Plus on one of the sites there were two backwards by mistake votes. So your 19 becomes 17 of 27 choosing jittered sound.

 

[bibo01]

I report my experience -

I was one of the 6 voters on NextHardware who chose A2, as I did not know the result before hand.
To be honest - I am not trying to "save" myself - in my first hearing I preferred A1; then the next day, listening with my partner, I preferred A2. Preferring A2, however, it left me with a doubt. In both cases, for example, I preferred the first song I heard. Then, when my friend Tom told me that in an ABX test he was not able to discern a difference through headphone, I realized it was not at all easy to distinguish the tracks in terms of jitter and no-jitter.

 

[lateboomer]

We did fall flat on our face. I have not unzip the file to confirm, but this is interesting and need to listen again why A2 sounds better to us. This is profound and throw a spanner into equipment assessment and sound tuning. Am I tone deaf and this hobby starts to get discouraging.

Anyway, thank you bibo01.

 

[BE718]

We did fall flat on our face. I have not unzip the file to confirm, but this is interesting and need to listen again why A2 sounds better to us. This is profound and throw a spanner into equipment assessment and sound tuning. Am I tone deaf and this hobby starts to get discouraging.

Anyway, thank you bibo01.

Im not sure we did fall flat on our face. If the test is representative (?!?), it is very interesting. I have certain opinions regarding the sound people like. They may well prefer certain distortions. As an example I have an Audiolab MDAC which has a bunch of different output filters. Now the one that people tend to prefer has a whole bunch of spuria, harmonics and intermodulation products up to around -110 dB with higher level signals.

1kHz & 1.1 kHz tones - 60dB Sharp filter (not generally liked - other filters apart from two of the optimal transient varieties have similar spectral characteristics)



Optimal Transient Filter (generally preferred)




Now some would argue in this instance the preference is due to the other filter characteristics, ie roll off and impulse response, but I don't think this is the case. There are 3 optimal transient filters with the same roll off and impulse response. one of them doesn't suffer the spuria and is not generally liked as much as the other two.

This even seems to imply that the distortions are not inherent to the filter type, makes me wonder if they are actually deliberate ! Makes it sound more "analogue".

 

[lateboomer]

But it still leaves a bad taste in the mouth. I always believe jitter and SQ are as clear as black and white.

 

[esldude]

But it still leaves a bad taste in the mouth. I always believe jitter and SQ are as clear as black and white.

I think you are assuming like many do, that audiophiles always prefer fidelity. That if jitter was enough to be audible then the no jitter file would get picked the most. I don't at all think audiophiles always prefer fidelity when there is a genuine difference. Many prefer something quite far from fidelity. The popularity of the LP and vacuum tube gear are two good examples. Some become quite heated in arguing that in some undefinable, unknowable way these lower fi options are in fact higher fidelity and better. It is more likely true that some just prefer that particular sound.

 

[BE718]

Very much agree with the above.

 

[lateboomer]

bibo01, do you think is it good to have another round of poll with same dacs but different song to choose the less jitter one? So by theory since we know A2 with more jitter and specific sound signature, we should able to score better, is it not?

 

[bibo01]

bibo01, do you think is it good to have another round of poll with same dacs but different song to choose the less jitter one? So by theory since we know A2 with more jitter and specific sound signature, we should able to score better, is it not?

I will ask Tom Gefrusti if he can prepare other two files.
I know that he is preparing a test where a cheap DAC + filter is indistinguishable from an Emm Labs

 

[bibo01]

bibo01, do you think is it good to have another round of poll with same dacs but different song to choose the less jitter one? So by theory since we know A2 with more jitter and specific sound signature, we should able to score better, is it not?

Tom is willing to have another round of poll. No problem.
It would ideal if you could supply a good analog track of your choice. Could you please upload it somewhere?
...or at least suggest a track you know.

------------------------

In the meantime I show you a zoomed picture of a NTD (please look in the Regen thread for instructions) applied to the two tracks - not much difference:

 

[lateboomer]

Hmmm....I think the first choice of music was very good because they were acoustic instruments. So follow that pattern I think good test records e.g. from Chesky will do, maybe music with guitar, sax or piano, with vocal is a plus bonus.

I also would like to suggest for those expert listeners who got the first poll result right refrain from voting until the last day. It is like in some car racing, the top qualifiers have to carry more weight or start at the back. This is to disadvantage them from getting lucky and prevent us from influenced by their comment. Just my 2 cents.

 

[bibo01]

Hmmm....I think the first choice of music was very good because they were acoustic instruments. So follow that pattern I think good test records e.g. from Chesky will do, maybe music with guitar, sax or piano, with vocal is a plus bonus.

I also would like to suggest for those expert listeners who got the first poll result right refrain from voting until the last day. It is like in some car racing, the top qualifiers have to carry more weight or start at the back. This is to disadvantage them from getting lucky and prevent us from influenced by their comment. Just my 2 cents.

OK, so you do not have a track of your preference. I will pass your Chesky suggestion on.

People can vote secretly - I think I can make the results "not visible" - and make comments without saying what they voted.

 

[lateboomer]

If up to me, I will use 4 analog tracks in the poll. 2 of it are based on same flute and harpsichord like the first poll but different song, and another 2 tracks are based on more familiar instruments like guitar, sax or piano.

So in this way, by using the first poll as reference for us, this 2nd poll will work like ABX test, at the same time also can test to identify dac with less jitter. Two birds in one stone. So we have to choose which 2 tracks have less jitter and total combinations of result are 6 to choose from.

Then we can eliminate a lot of fluke results.

 

[bibo01]

It would be too complicated for a poll, but I will pass your suggestions for the type of instruments.

 

[Orb]

Im not sure we did fall flat on our face. If the test is representative (?!?), it is very interesting. I have certain opinions regarding the sound people like. They may well prefer certain distortions. As an example I have an Audiolab MDAC which has a bunch of different output filters. Now the one that people tend to prefer has a whole bunch of spuria, harmonics and intermodulation products up to around -110 dB with higher level signals.

1kHz & 1.1 kHz tones - 60dB Sharp filter (not generally liked - other filters apart from two of the optimal transient varieties have similar spectral characteristics)


Optimal Transient Filter (generally preferred)


Now some would argue in this instance the preference is due to the other filter characteristics, ie roll off and impulse response, but I don't think this is the case. There are 3 optimal transient filters with the same roll off and impulse response. one of them doesn't suffer the spuria and is not generally liked as much as the other two.

This even seems to imply that the distortions are not inherent to the filter type, makes me wonder if they are actually deliberate ! Makes it sound more "analogue".

Well worth noting that optimal will also have weaker alias rejection and causing imaging, unfortunately there is no hard and fast rule how weak/strong this needs to be before influencing a listener and whether good or bad.
Although worth noting NOS DACs do not remove any images (cannot really use this to correlate as a reason why NOS DACs sound good but an indication we may be able to handle this factor quite well in terms of a dissonance type effect and this is the most extreme example of imaging/alias effect).
Also there is greater IMD with weaker alias rejection - this will definitely be more noticeable if using two tones when test measuring but again comes down to the reconstruction filter design and implementation.

Cheers
Orb

 

[BE718]

Well worth noting that optimal will also have weaker alias rejection and causing imaging, unfortunately there is no hard and fast rule how weak/strong this needs to be before influencing a listener and whether good or bad.
Although worth noting NOS DACs do not remove any images (cannot really use this to correlate as a reason why NOS DACs sound good but an indication we may be able to handle this factor quite well in terms of a dissonance type effect and this is the most extreme example of imaging/alias effect).
Also there is greater IMD with weaker alias rejection - this will definitely be more noticeable if using two tones when test measuring but again comes down to the reconstruction filter design and implementation.

Cheers
Orb

Yes, aware of the potential affects of the filter. Interestingly the designer is quite cagey about the details of the 3 OT filters. As mentioned one simply doesn't suffer the spuria.

Also whilst many people seem to like these filters, personally I don't. I found them sounding less clean and rolled off, which when I measured them I subsequently found they were indeed just that.

 

[853guy]

I wish I could offer a more articulate and eloquent explanation for why I preferred the jittery track over the non-jittery one, but I can't. I wish there was some way of measuring my perceptual experience in correlation with the way I voted, but there isn't. Would I have voted the same had it been a different track, with a different listening chain (laptop jack, headphones), in a different environment (kitchen), on a different day? Perhaps, but we'll never know. Suffice to say, if the experience I had with the more-jittery track is anything to go by, it perhaps at least explains why I still consider vinyl and tape to be high-water marks for music replay, knowing there's still so much to be gained from further improvements in the reproduction of both those mediums.

 

[Groucho]

I wish I could offer a more articulate and eloquent explanation for why I preferred the jittery track over the non-jittery one

I started off convinced I could hear the difference (and correctly - hey! I must have Golden Ears) but then I started playing the files randomly without knowing the filenames. I couldn't, in fact, identify the difference consistently at all. I had latched onto an imaginary difference and was from then on reinforcing that bias every time I played the two files 'sighted'. The fact that I didn't know which was which was irrelevant: I made my own bias based on my first guess (which was always going to be 50% chance) and then stuck to it based on the file name. If the order of the two files had been different, I might have guessed the other way.

Should anyone be expected to hear a difference anyway? I would say no: jitter is a classic case of "masking". The jitter 'noise' is directly proportional to the amplitude of the desired signal - zero signal : zero jitter 'noise'; it will always be dwarfed by a much louder signal. Maybe it becomes more audible against certain signals where the noise's spectrum is different from the desired signal, but unless the jitter is massive, it will still be very, very quiet.

 

[bibo01]

I published test Part II and Part III