Synergy?

[Phelonious Ponk]

Well, happily I do not want to carry serious listening!

I have to say I do not care about this type of detail. My listening is mostly driven by the performance. I do not care if it is a Stradivarius or a Guarneri, a Steinway or Bösendorfer. But I want to feel the energy of the performers, their complicity, the beauty of a voice, the power of a finale in a great auditorium and the intimacy of a chamber performance. I want to be able to feel the performance of Claudio Arrau, feel all these subjective thinks such as the rhythm, the music growing from a black silence, the grandiosity due to the back reflections, the emotion carried in a very pure treble, the drama in the end of the Rigoletto, the beauty of the voices in Monteverdi. And, sorry, but I do not believe that training me to listen to recording artifacts or defects, or evaluate frequency response effects by listening will increase my listening pleasure.

Enthusiastically agreed.

Exceptionally I can not resist to serious listening activities - I love to look for the horseflies in Gregorio Paniagua "La Folia" recording to test the resolution of a system!

I would much rather listen for the lovely texture of the aging diva's voice, personally. Different strokes...

Tim

 

[Phelonious Ponk]

Interesting.

For me, I want to hear the details, the clarity, each sound distinct. Width, depth, holography, soundstage, etc, all take a back seat for me. I dont suspend belief nor care to. Resolution is what turns me on. Unless, of course, when I want to jazz stuff up, then out comes the SET or the image expander, etc. But if I could only have one thing, its the resolution that drives me.

Tom

I agree with this 100%. Detail, and its natural subset, imaging, not to be confused with sound stage, which can actually seem to grow as detail shrinks, according to some descriptions. I also think realistic tonality runs hand in hand with the resolution of detail.

Tim

 

[jkeny]

I don't think science and objectivity supports a stance that says 10 picoseconds difference is audible. At least I don't know how to get there. The difference between 50 and 60 psec peak to peak jitter is -114 dB and -116 dB distortion products. If I constructed such an experience artificially, I don't think people will be able to hear that difference.

As far as I am concerned, once someone gets the jitter down to tens of picoseconds the job is done. ........

Amir, just to go back to something & update it with recent reading. You have stated in the past that 20bits is the target bit depth needed to be sure of fidelity. This analysis shows that 8pS of jitter is thus the audible threshold level for 20bit resolution. So I guess in the same spirit of WBF, 8pS should be the standard?

A better estimate of the audible jitter threshold can be obtained by examining the jitter error sequence, and assuming that it will be inaudible if below the level of quantization noise present in the system for any possible excitation frequency. Lidbetter (footnote 25) thus arrives at a value of 120 picoseconds for a 16-bit, 100% sample DAC, and an incredibly low 8ps for 20-bit system. Shelton (footnote 26), Fourré (footnote 27), Harris (footnote 28), van Willenswaard (footnote 29), and the recommendations embodied in AES11-1991 (footnote 30) all quote similar values.

http://www.stereophile.com/content/bits-bits-page-7 I think it might have been an article you referenced some graphs from when arguing for the 20bit threshold of optimal bit depth.

 

[amirm]

Amir, just to go back to something & update it with recent reading. You have stated in the past that 20bits is the target bit depth needed to be sure of fidelity. This analysis shows that 8pS of jitter is thus the audible threshold level for 20bit resolution. So I guess in the same spirit of WBF, 8pS should be the standard?

There is a difference between what you are asking here and what you quoted from me. I said that the difference between -114 and -116 db in distortion can't be audible and certainly not with the description of large differences.

As to the criteria, for a 20 Khz bandwidth channel, 40 picoseconds peak to peak gives us -120 db distortion peaks. While we can't quite equate distortion with noise level, the latter for 20 bits is 120 dB. 40 psec is the tens of picoseconds region of "tens of picosecdonds" I mentioned in my post as being pretty good.

http://www.stereophile.com/content/bits-bits-page-7 I think it might have been an article you referenced some graphs from when arguing for the 20bit threshold of optimal bit depth.

My source is the same as that article: the Bob Stuart AES paper and to some extent, that of Hawksford and Dunn.

 

[jkeny]

There is a difference between what you are asking here and what you quoted from me. I said that the difference between -114 and -116 db in distortion can't be audible and certainly not with the description of large differences.

As to the criteria, for a 20 Khz bandwidth channel, 40 picoseconds peak to peak gives us -120 db distortion peaks. While we can't quite equate distortion with noise level, the latter for 20 bits is 120 dB. 40 psec is the tens of picoseconds region of "tens of picosecdonds" I mentioned in my post as being pretty good.


My source is the same as that article: the Bob Stuart AES paper and to some extent, that of Hawksford and Dunn.

I highlighted the most important statement in what you said - we are not talking about noise (random Gaussian jitter) but about distortion (data dependent jitter - DDJ) which was what was being discussed. In fact I would say that the audible level for Random jitter is in nano seconds & not pico seconds!!

 

[amirm]

I highlighted the most important statement in what you said - we are not talking about noise (random Gaussian jitter) but about distortion (data dependent jitter - DDJ) which was what was being discussed. In fact I would say that the audible level for Random jitter is in nano seconds & not pico seconds!!

My criteria remains for correlated jitter, not random. I don't even bother worrying about random jitter. In the context of correlated jitter, I believe once the distortion spikes are in -120 db area, we are golden regardless of audibility difference between tone and noise.

Note that my feelings in this regard are a bit different than what should be in a recording. There, I simply advocate not degrading whatever is there on the path to the consumer. I see no reason to reduce the fidelity for the sake of doing it, now that we are free of CD format restrictions. A recording is made once and let's make it the best it can be. Once there, folks can draw the line where they like as far as reproduction. I set no maximum performance level there but rather the minimum. I say that let's make sure we get -96 dB. I think I can demonstrate benefits of that. I don't know how set that minimum at -116 or whatever got us in this argument.

 

[jkeny]

My criteria remains for correlated jitter, not random. I don't even bother worrying about random jitter. In the context of correlated jitter, I believe once the distortion spikes are in -120 db area, we are golden regardless of audibility difference between tone and noise.

Note that my feelings in this regard are a bit different than what should be in a recording. There, I simply advocate not degrading whatever is there on the path to the consumer. I see no reason to reduce the fidelity for the sake of doing it, now that we are free of CD format restrictions. A recording is made once and let's make it the best it can be. Once there, folks can draw the line where they like as far as reproduction. I set no maximum performance level there but rather the minimum. I say that let's make sure we get -96 dB. I think I can demonstrate benefits of that. I don't know how set that minimum at -116 or whatever got us in this argument.

OK, you apply a different criteria for recording to playback, I understand that. But are you disagreeing with this quote from the Dunn & Hawksford paper & why?

A better estimate of the audible jitter threshold can be obtained by examining the jitter error sequence and assuming that it will be inaudible if below the level of quantisation noise present in the system for any possible excitation frequency. Lidbetter [20] thus arrives at a value of 120 ps for a 16 bit 100% sample DAC and an incredibly low 8 ps for 20 bit system. Shelton [21], Fourre [22], Harris [8], van Willenswaard [7] and the recommendations embodied in AESll-1991 [23] ail quote similar values.

 

[amirm]

OK, you apply a different criteria for recording to playback, I understand that.

That was not quite the point I was making but let's move on.

But are you disagreeing with this quote from the Dunn & Hawksford paper & why?

That is not the view from the authors but quoting the summary of the paper referenced. They go on and perform their own analysis:

"Fig 35(a) shows the worst case 100% DAC jitter error resulting from a 22 KHz audio signal and 18.5 KHz jitter -- only 20 ps peak jitter is required for audibility."

That is essentially the same criteria I mentioned earlier in my 40 ps number with the difference being that they are assuming 22 KHz bandwidth while I used 20 Khz.

 

[jkeny]

That is not the view from the authors but quoting the summary of the paper referenced. They go on and perform their own analysis:

"Fig 35(a) shows the worst case 100% DAC jitter error resulting from a 22 KHz audio signal and 18.5 KHz jitter -- only 20 ps peak jitter is required for audibility."

That is essentially the same criteria I mentioned earlier in my 40 ps number with the difference being that they are assuming 22 KHz bandwidth while I used 20 Khz.

Yes but if quantisation noise is important enough that dither is applied to avoid it then why is jitter distortion at this level not important enough? If we were to just deal with audibility then why bother with dither down at this level?

 

[amirm]

Yes but if quantisation noise is important enough that dither is applied to avoid it then why is jitter distortion at this level not important enough? If we were to just deal with audibility then why bother with dither down at this level?

Distortion is additive so one thing doesn't justify the other. And further, getting the channel (distribution format) to be free with headroom even is free these days. Any argument against that is then silly. It would be like demanding that we add distortion to the music or else!

The decode/playback side is a different matter. I can show that getting down to 500 psec is "fee" so going below that is like above is like the above argument of living with distortion for the sake of it. Past that it gets progressively harder to demonstrate that.

 

[jkeny]

Distortion is additive so one thing doesn't justify the other.

So why add unnecessary distortion in the form of jitter even if you consider it is below the noise floor?

And further, getting the channel (distribution format) to be free with headroom even is free these days. Any argument against that is then silly. It would be like demanding that we add distortion to the music or else!

I don't follow what you are saying here?

The decode/playback side is a different matter. I can show that getting down to 500 psec is "fee" so going below that is like above is like the above argument of living with distortion for the sake of it. Past that it gets progressively harder to demonstrate that.

Again, I'm not sure what you're saying here?

 

[amirm]

So why add unnecessary distortion in the form of jitter even if you consider it is below the noise floor?

I am not aware of people using a jitter knob and adding that to their players. Whereas if they don't use the right knob to down sample the music, they would be adding distortion.

I don't follow what you are saying here?

You were contrasting actions taking in post production to that of playback (dither vs jitter). I explained that there is no cost to getting the absolute fidelity in the channel. The only way to mess it up is to use the aforementioned knob to add distortion on purpose.

Again, I'm not sure what you're saying here?

I thought it was pretty obvious. I can show you how even a cheap playback system can achieve better than 500 psec jitter. So the argument that this is all about expensive audiophile solutions does not apply here. As you go past that and start to argue for 10 or 20 picoseconds or whatever, it is hard to make the same argument.

My position in all of this is one of reasonableness. I don't support either extremes of this hobby. I go as far as I can defend things with a straight face. Past that, others need to go on the journey, not me . A lifetime of trying to optimize things within the means given has gotten me here.

 

[jkeny]

I am not aware of people using a jitter knob and adding that to their players. Whereas if they don't use the right knob to down sample the music, they would be adding distortion.


You were contrasting actions taking in post production to that of playback (dither vs jitter). I explained that there is no cost to getting the absolute fidelity in the channel. The only way to mess it up is to use the aforementioned knob to add distortion on purpose.

It's late & I'm obviously tired or thick but I find this too convoluted for me to follow. Maybe tomorrow when I look at it again, it will all be clear.

I thought it was pretty obvious. I can show you how even a cheap playback system can achieve better than 500 psec jitter. So the argument that this is all about expensive audiophile solutions does not apply here. As you go past that and start to argue for 10 or 20 picoseconds or whatever, it is hard to make the same argument.

My position in all of this is one of reasonableness. I don't support either extremes of this hobby. I go as far as I can defend things with a straight face. Past that, others need to go on the journey, not me . A lifetime of trying to optimize things within the means given has gotten me here.

OK, we agree to differ & I'm on my own here

 

[NorthStar]

----I'm here too, and listening (reading) e.v.e.r.y.t.h.i.n.g. :b

 

[Gregadd]

The Joker commenting on Batman: "Where does he get those gret toys?"

Comenting on AMIRs graphs.