DAC Linearity & Percieved Detail

I stumbled upon this, and while it was an interesting read, the comments got unnecessarily contentious, so it left me with questions that I felt might be better presented here.

Basically, the author, John E. Johnson, with input from Nelson Pass and John Rich, makes a bold statement. He believes that nonlinear response in a DAC might actually boost detail retrieval from the nether regions between the audible threshold and noise floor, thus improving the perception of detail, so long as the time domain of the signal does not get shifted too far out of whack. At least, that is my layman's impression of the idea.

Is this a thing? It seems reasonable to assume that a digital signal could be differentially boosted, thus presenting to the listener audible detail that would normally reside at amplitudes that would prevent it from having an impact on the listener. Is this what is meant by 'plankton'? Is a pleasantly non-linear response to low level input a factor in distinguishing highly musical, involving digital reproduction from a device that tests well but sounds shitty? Has this been discussed ad-nauseum in other threads but called something else that eluded my searches?

 

Probably not. Objectively, crest factors of many digital musics are only between 8-15db for compressed pop musics, and 20-30 db (more or less) for uncompressed acoustic tracks. So, I can't really get to the idea effective bits and/or gain linearity at the lower end matter in perceived details. Academically interesting though.

And subjectively, I know a bunch of dacs that measured excellent in gain linearity but cannot resolve dynamics and details at least to my ears (topping dacs may be a good example).

So my take about such measurements is the relationship and interpretation is rather complicated than straightforward/comprehensive, particularly regarding real-human perceptions.

 

By the time r2r DACs go non-linear you're already below the audible threshold. This can be easily tested yourself. Create a 1khz sine wave test file -90dbfs, you'll never hear it. I can't hear anything quieter than -80dbfs and that's late at night with my ear straight up to my horn speakers. At my listening position it's still dead silent.

 

Do I understand this argument correctly? : He is saying that a consistent change to the signal (i.e. "non linear" - in other words a distortion), in particular at low amplitude (i.e. does not occur or is inaudible to the listener if the signal is at a higher amplitude?) is "boosting" detail? So distortion as detail?

 

It is almost as if he is positing some sort of distortion as dither...or something...

 

So his contention that the measured non-linear response was responsible for a more pleasing sound and better low-level detail is, at best, unfounded, and at worst, a complete misapprehension and totally misleading to those of us who are kinda dumb at this stuff.

I guess what interested me is the idea that maybe there was a glimmer of hope for a way to measure, objectively, the relative level of audible detail retrieval from one component to the next. Subjective input is fine, but it would sure be nice to point to a number on a report or a graph that indicates superior performance in some of these areas where subjective statements seem to rule the day. "OMG, the plankton from this DAC!" "OMG, this amp produces such a wide soundstage!" "OMG, the dynamics are just soooo immersive!" Meanwhile, the next reviewer is trumpeting the superior performance of his\her favorite component brand, which obviously sounds so much better.

I know, I know....just listen to it. I hear that all the time. Problem is, when I cover my eyes, all bets go out the window, just the same as everyone else. Does that make me a wanna-be objectivist?

 

Kinda understand you. I call it "the trap of audio-objectivism". Once thought quite similarly, then sold all my rigs and just sticked to chi-fi gears which measured well (knew topping in this period). I thought I heard as equivalent sound as uber gears. But I became not to listen musics as much. After several years, learned a very important lesson in the end: "trust your ears".

Now I think in this hobby finding the "right" subjective reviewers whose statements can predict my preference/satisfaction (this actually needs careful research to find out where the reviewer came from, how he/she expresses things, and whether he/she is just an idiot) is much more doable and important than playing with technical concepts which I only understand superficially (like gain linearity in this case).

I am afraid that sbaf is not a place filled with active discussions of things which you might want, although there are many experts. Hydrogenaudio might be more suitable place imo.

EDIT: attach one graph as an example (copied from this thread). This dac measured well. Excellent result from your measure/theory, but my subjective experience was terrible with this one.

 

Oh, I trust my ears, as long as they can't see. The fact that sighted listening is flawed is so well documented that it does not even bear discussion. That will not stop me from carrying on ad-nauseum about my perceptions of a listening experience, nor should it. BUT, it would be nice to put some metrics to the ephemeral qualities we all love to yammer on about. Not trying to rain on the subjectivist parade, just hoping to add some data that might dispel or reinforce the almost religious nature of the conversation.

Sounds like this particular bit of errata failed to pass muster.

 

Indeed you added NOTHING. Gain linearity is often measured and reported in a number of measurement reviews. We also know there are a number of measuring guys who mislead or are misled (Hint: google jude or amir) - I am also suspecting the author of your referred article is one of them. There is also consensus that understanding dac or amp measurements are not straightforward as you think (this is well documented, too).

Underlined parts are particularly disgusting. Note that one cannot add valuable points unless he/she knows related things well enough with a good amount of practical experience, particularly regarding measurements.

 

What’s your take on Bob Katz’s recent piece on small amounts of distortion sounding nice? https://www.innerfidelity.com/content/katzs-corner-episode-25-adventures-distortion

 

I have been thinking about that article for a few days. Right now, my conclusion is that Bob likes a bit of Tube euphonics to go with his SS (so do I). Is it possible that his experiment reveals that in some cases, a little 2nd order can mask unpleasant non-linear? I don't know the answer...

 

Totally agree.

The author had an interesting (if flawed) premise, and given the input of some respected industry names, it seemed an odd departure from the little I think I understand, and therefore worthy of discussion. If my opinion of the nature of subjective experience disgusts you, then perhaps we can just agree to disagree on that subject? We're all after the same thing, at the end of the day....a pleasant listening experience with music reproduction, and the continued better understanding of same.