I don't want to beat a dead horse but I'm bugged.

Great post!

I will only add that we cannot know how to measure and how to interpret these measures correctly if we dont have a correct model of human hearing at the end......

That was the subject of my last 3 posts here.... With some interesting  scientific articles about hearing ....

@deludedaudiophile

I came up with the 95% missing using historic precedents for scientific discoveries. (Throughout history the narrow minds all firmly believed that everything that could be invented was already invented, and now even a 6 year old kid knows it better.) The practices for audio gear measurements are relatively new (just a few decades old). In even 50 years, our practices will be proven as massively inaccurate, and in general quite useless as it probably covers about 5% of what our children’s children will count as measurements that point towards sound quality perception. Although that will be in the future, yet it does not detract from the reality that our current practices are in their infancy. To think we know everything, and we have discovered all the secrets to sound and audio gear is the only sure bet to loose. Doing a google search will do no good now, but will help in 50 years. Also, if google search would answer deep questions on audio measurements, we would not have this discussion, and everyone would be at perfect agreement.

(BTW the 95% is just a symbolic value, please do not start a thread on whether it is 92% or 96.786734% exactly, or it’s truly 57.4%... only time will tell, and although our view will change decade from decade, but the reality will be still the same: today we have a very limited concept of how to measure audio parameters to reflect on sound quality.)

So, a few examples on issues with current standard measurement practices that I know of, I will take only amplifiers for now:

*all parameters tested on non-inductive perfectly passive extremely simplistic loads, while the loudspeakers are highly complex live loads affected by the room.

*Only additive distortion is measured, subtractive distortion is not.

*Change of THD in function of output level and frequency are no paid attention to, while these are strong determiners in relation whether the sound is perceived as natural VS manufactured.

*Amplifier behavior is tested with constantly repetitive primitive signals, while the music output is a highly variable extremely complex waveform.

*It is not examined how an amplifier deals with small signals following a large pulse at the frequency extremes.

 

I mentioned the names of known and proven audio authorities in my initial post, because they have the answers you want from me, but I have no credibility in your eyes, so it’s a waste from me to yap around. Do not believe me, as you do not know who I am, and that’s fine.

I just humbly point out to you (again), to listen to interviews with the fathers of audio measurements and high end industry and hear what they have to say. Thank you for the chat. I hope I have answered your concerns. What I wrote might be completely irrelevant to your quest, and it’s quite likely that you have specific experiences that point you in the direction you want to go, where you will find fulfillment and purpose.

However, ignoring the experts on audio measurements will quite likely lead to a more protracted learning curve than what you are looking for. I wish you success and luck!

@realworldaudio , I am not an EE, but my I have an advanced physics degree, and have worked in semiconductors, batteries, and development and measurement of those a long time. I have gotten pretty good with metrology out of necessity. Hence I don’t claim to be an expert, but I think I have a good grasp of what is being communicated in the measurements:

*all parameters tested on non-inductive perfectly passive extremely simplistic loads, while the loudspeakers are highly complex live loads affected by the room

Stereophile specifically mentions they use for some of their tests a synthetic speaker load that models a real world speaker. This appears to negate some of the above statement. Testing into 4 ohm is standard. Testing in 2 ohms seems common. This will provide insight into more reactive loads. Testing with worse case synthetic loads is common and harsher than real world conditions. For characterizing semiconductor devices we test with synthetic loads to find the "corners" for stability.

*Only additive distortion is measured, subtractive distortion is not.

I am not sure exactly what you mean by subtractive distortion. Are you stating that the interaction of a particular amplifier with a particular speaker may result in lower overall distortion? This seems possible. I will note that @atmasphere who seems to know his stuff stated most (not all) speakers are designed to be driven with a voltage source which may negate the advantage you may perceive for most listeners. This would be dependent exclusively on the load (speaker) so I don’t see how this could be tested.

*Change of THD in function of output level and frequency are no paid attention to, while these are strong determiners in relation whether the sound is perceived as natural VS manufactured.

This is purely false. Read any of the more recent reviews on ASR. THD and SINAD is tested from very low power to very high power across a range of frequencies from I think 20Hz to 15KHz. I am too lazy to go verify the exact frequencies used.

*Amplifier behavior is tested with constantly repetitive primitive signals, while the music output is a highly variable extremely complex waveform.

This is also purely false. ASR tests with a 32 tone IM signal. This is 32 tones from low frequency to high frequency. That would result in a signal that is complex and varying in amplitude as the frequencies add together.

*It is not examined how an amplifier deals with small signals following a large pulse at the frequency extremes.

I will not call this false, but I think you are not interpreting what the other measurements will accomplish. The 32 tone IM signal will vary from large to small. The THD stimulation also transitions from a very small level to very large. If the measurement is -100db in both cases then that would also be the case for the special condition you are theorizing.

What may be missing is testing if the distortion rises under continued heavy load causing device heating. I do not know if that is a valid real world condition.

I have listened to Nelson Pass. He strikes me as very much a heavy measurements guy. He may tune intentional artifacts in his designs, but what I have read and what I have been on Youtube indicates he is very much measurement oriented.

@realworldaudio Thank you for stating what should be obvious. Of course some believe we've already discovered all there is to be discovered. Why not just shut down any research into audio reproduction and human sensory perception, we already have all the measurement protocols and tools needed to prove absolutes, and differences in individual human sensory perceptivity is of no concern.

Watch out for the audio authoritarians, they'll be sure no Toto can open the curtain.

Are you any better @sns , insisting on magic because you won't accept your own infallible and easily influenced hearing?  I replied to @realworldaudio's post, objectively, with no idea or care for his qualifications. Some of the things he said are wrong. Some of the things he says seem relevant, and some are likely outside the scope or even capability of any viable test regimen but does not make them irrelevant, just impractical.

Who would consider this a good frequency response?

It’s the Harman curve. Research showed this to be the frequency response for headphones the majority of people like best.

It’s curves like this (there are several variants) that headphone designs nowadays target. Maybe we should do something similar for Amps plus speakers.

Tone control is mostly stripped off of hight end amps, let alone the good old 'loudness' switch. A multiband equalizer is a very nice piece of equipment ... strangely enough seldom used in high end hifi. I wonder why?