Sonic Transparency and Listener Bias: What We Hear

Very interesting article.

Yeah. We tend to get lost in the weeds in discussions here. That guy's got a gift for making describing the technical in non-technical language.

For me this was good, delicious Motherhood and Apple Pie. Nothing new, but well-known stuff stated in a friendly way.

One of the most amusing parts of his comment about obvious technical incompetence at high end web sites, with comments about a certani power amp http://www.high-endaudio.com/RECENT.html#Mar as an example.

The high end expert says: Every listener described the Halcro as being 'unlistenable', justifiying it with this claim: "To give you an example: the Halcro DM58 amplifier measures .007%THD at 2 watts, whereas at 1/1000th watts, THD measures 8.9%! By contrast a Wavac SH833 measures .57%THD at 2 watts and 1.6%THD at 1/1000th watts. The tube Wavac exhibits significantly lower THD at low signal levels by orders of magnitude than the Halcro. I have measured numerous amplifiers, both solid state and tube, which I will provide to you as well as any other information you may want pertaining to this work."

There is an independent test of the Halcro posted at http://www.stereophile.com/content/h...rements-part-2. which includes the following test results:

In fact the above shows that all harmonics are 105 dB or more down.



Shows no actual evidence of harmonics increasing dramatically at low levels. Instead it shows the results of a constant noise floor being compared to a varying signal level.

It is too bad that John Atkinson doesn't do standard dynamic range testing of power amps because it gives more direct evidence about performance at low listening levels, but John appears to be a very old-school guy when it comes to measurements.

Nice article and the BBC video was very interesting.

Good article. Like Arny, not much new to me but well presented and worth bookmarking to point others to.

In case anyone hasn't seen these three clips, they are great to show the effect of presentation on perception. Damn funny too.

Bottled Water
The best
No comment

Thanks for the article. The author clearly spent a goodly amount of time on it.

IMO comparing wine to audio is a mistake. Sure they both suffer from preconceptions. But the fact is most wine (or beer - my beverage of choice) does taste different.
Most good audio equipment sounds very nearly the same.
I'm not sure THD is a good measure. I think some people like tubes because most of the distortion is even harmonics (kind of like what most instruments make anyway) whereas most transistors make both even and odd harmonics.

I don't have a single piece of tube gear. But I have thought about adding a tube distortion on purpose - I don't think you can do that with an eq. Maybe you can with DSP. Some music I think might benefit by adding some nice even order harmonics (perhaps removed in a studio that uses too much eq). I'd probably only use it on things I thought were badly recorded, and then only very rarely. I'd like a simple switch - add .2, .5, 1 even order harmonic distortion. If nothing else it would be interesting to see if I could actually hear the difference.

Perhaps I should not comment yet - I haven't made it through the whole article - but will to see what happens :cheers:

That's inaccurate description of tube and transistor audio gears. Both can have more or less even harmonics depending on the circuit design type (single ended vs push-pull).

I'd like to learn more about this.

For starter:
https://en.wikipedia.org/wiki/Amplif...lifier_classes
https://passlabs.com/products/amplifiers
http://diyaudioprojects.com/Schemati...-Schematic.htm

After reading NwAvGuy's a posted a little heads-up about it on my audio blog. Here's the passage that expands on what you wrote.

Good points. Of course speakers and rooms sound incredibly different, perhaps more different than different versions of the same wine. I think this may be why so many audiophiles think all components sound different - they hear so many of the different options in different rooms and with different speakers.


With electronics THD isn't a good measure because as used, just about everything has either no audible distortion or distortion so low that it is really hard to distinguish.


Actually transistor amplifiers being almost universally push-pull, have vanishingly low amounts of even distortion. Clipping is all odd order distortion. Tubed amps have odd order distortion when they clip, and often have lots of it below clipping. Since they typically have so much more nonlinear distortion, they almost always have more odd order distortion than a comparable SS amp
.

The story about people preferring even order distortion is probably at best a half-truth. Even order distortion is more likely to be masked by the ear (because most even order distortion is second order and thus closer to the fundamental and thus under its masking curve) and also by the natural sounds of musical instruments which are generally heavy on even order distortion.

Trouble is that THD is caused by nonlinear distortion which can't help but produce IM products as soon as two or more tones are present, which happens with just about all music. IM is primarily aharmonic which means that there are few naturally-produced tones in the music to mask it. Not only aren't IM products not masked and covered by the music, but they are generally at odds with the chord structure of the music. They just don't fit, and the ear knows that and gets unhappy.


There is a way to add controlled amounts of various orders of nonlinear distortion in a way that is similar to what an equalizer does for timbre. It can be done with some DAW software (CEP/Audition for example), but it can also be done with a DSP in real time.

The problem with using natural distortion generators is that they have one profile of distortion generation, which you can vary in intensity by controlling the size of the signal passing through, but you can't change its internal order in ways that would allow you to spin knobs and get the distortion of your dreams or favorite SET amp.


For sure you can with a DSP or a PC. Just write a little code in C...


Eq doesn't remove the effects of nonlinear distortion unless you are very, very intentional about what you do. Even if you are intentional most studio eqs can't do this well.

The best way to remove the effects of nonlinear distortion would be a nonlinear compensator, which is the technical name of what I was describing above. Nonlinear distortion can be cancelled out if you know what it is.


I think you'd quickly find out about how much you can't adding IM when you add harmonics, and how much you don't like IM. IME while tubed equipment can have scary amounts of nonlinear distortion, that isn't their most audible effect.

The most audible effect of tubed electronics is IME the poor frequency response, particularly in power amps driving loudspeaker or loudspeaker-like loads. Read a Stereophile review of tubed amps. Know that Atkinson's speaker simulator is easier on amps than many if not most speakers. I studied the kind of loads many of the speakers he's tested, and came up with a speaker simulator that was about twice as stressful.

Tubed preamps are actually pretty clean, especially if you keep them away from SS amps with low input impedances and long cables. I have a CJ and its not at all that different sounding than my SS gear.


Actually, that particular device has been built and its schematic is at:

http://home.provide.net/~djcarlst/abx_grun.htm (parts values for gvein amounts of THD at URL)
and looks like this:

The short answer is that a well-balanced push-pull amplifier inherently has very low even order distortion because the pushing and pulling automagically cancels it out.

True for tubes, true for SS. The idea that SS amps built in the last 3-4 decades have more odd order distortion than a compable tubed amp is an audiophile myth, as I explained above.