Gene got the Ban Hammer!

[killdozzer]

" ... When I discover new facts, I change my opinion. What do you do sir? ..."

-John Kenneth Gailbraith

Auuuu... I have a slight feeling not many people see you as someone who should post this.

 

[Matthew J Poes]

There are a number of reasons our brain can detect sounds differently than a measurement mic. Many of the ones noted in these examples come from our brains own dsp.

However one of the most important differences is that we hear with Two ears that have a cardioid pattern each. They combine to create something that is a little more hypercardioid like.

https://pdfs.semanticscholar.org/bb42/962552aa2f47c3cf4e511843c9b22d1c6740.pdf
Here is an older article that looked into this and is freely available.

Look at the right ear polar patterns


If you mirror those you have the binaural polar pattern of our hearing. However we can never forget that while the total of the mirror image is our polar pattern, it was created by two discrete sound detection devices.

This has a few important implications:
1)our hearing is forward biased. Not only can we detect direction better, we can focus on sounds better (and our bandwidth is wider-we hear higher frequencies).
2)there is a shadow at the back/sides of our head which are not areas of deafness but areas of reduced sensitivity. This allows us to tell when noises are behind us but also to block them out when needing to focus on the direction of sounds.
3)our hearing is binaural. We can’t ignore that fact when thinking about how we hear or how to replicate that with a mic. If you want to know what you hear when you sit in your listening position, you really need binaural mic (and even that isn’t right because a microphone element still doesn’t work like an ear).

Something that article doesn’t address is the interesting means by which the brains can detect direction, especially elevation by the frequency pattern that emerges or changes as an object moves. We actually have the ability to hear some directional effects via mono cues. This is essentially a kind of dsp in our brain but is related to resonances that occur in our ear. A mic can’t do that even with intellegent dsp. Maybe some day, but not today.

This is why attempting to measure what you hear is a fools errand. No amount of measurements will equal that. You are just creating an approximation.

 

[KEW]

Consumer Reports always had an interesting approach.
As I understand it they would measure the output and compare it to the signal to evaluate accuracy. It resulted in a single number (maybe % accurate?).
It may be very wrong to combine errors of distortion in the same category as, say, a premature roll-off of high frequencies, but I always wondered why no other labs adapted that approach...if anyone has familiarity!

 

[Matthew J Poes]

Consumer Reports always had an interesting approach.
As I understand it they would measure the output and compare it to the signal to evaluate accuracy. It resulted in a single number (maybe % accurate?).
It may be very wrong to combine errors of distortion in the same category as, say, a premature roll-off of high frequencies, but I always wondered why no other labs adapted that approach...if anyone has familiarity!

That is almost literally how you derive an impulse response. You convolve a reversed recording of a system with the original test signal and out comes the impulse response. If you did that with the original test signal only you would get a flat line equal to the bandwidth of the test signal, no distortion, and perfect decay. Any devatiaoth to that is the distortion caused by the reproduction system and room and is what causes all of the distortions (linear and non-linear) you see in the resultant impulse response.

What isn’t common (because it has little relationship to what matters most, what we hear) is trying to reduce that to a single unweighted percentage of error. The reason is because all systems will always introduce some amount of errors. Rooms and speakers introduce a lot of errors. Trying to treat all errors as equal would negate the fact that some errors are in fact inaudible and certain ones more objectionable than others. There is not likely enough research to be able to even weight and score a product by the error of its impulse response and fully capture our perception of quality. I would guess Harman has come the closest and it’s a more nuanced process than simply looking at the error in the output relative to the input.

 

[Matthew J Poes]

Oh and one last thought. A single impulse response would also fail to capture all aspects of a system. Harman rightly makes a big deal about distortions of the polar response which are essentially many measurements taken at different angle. Not all angles matter as much as others. More distortion at the extremes is ok, for example. So even trying to average the error score across many measurements wouldn’t accurately capture how a system sounds.

That doesn’t mean this can’t be done. It just means it’s a more complex problem.

 

[lovinthehd]

Another thought is why does it bother people so much to consider the use of measurement gear? Analyze isn't something I particularly think of as an exercise by ear but rather instrumentation, too....but maybe that's just me. No one was suggesting not to still use one's ears....and many of the responses in that original thread were from people who want to only use their ears exclusively....

 

[Matthew J Poes]

Another thought is why does it bother people so much to consider the use of measurement gear? Analyze isn't something I particularly think of as an exercise by ear but rather instrumentation, too....but maybe that's just me. No one was suggesting not to still use one's ears....and many of the responses in that original thread were from people who want to only use their ears exclusively....

I’ll be honest I only read what Gene wrote and responded here because James sent me the link. So I may be speaking out of place here.

Measurement mics don’t hear what we hear, but that doesn’t mean they are an invalid tool for scientific or even setup purposes. Most people’s hearing isn’t refined enough (I.e. trained) to use in place of a mic. Even if it was, it would likely be a slower process. That doesn’t mean a person can’t tune a system by ear successfully. The bigger issue is that because people falsely assume that a mic hears what our ears hear they over rely on the measurements. The worst offenders are those who take a single measurement and then eq it flat across the whole band. However it’s not much better to just take a handful of measurements, average then, and do the same. A small amount of eq below 500hz to knock down big peaks taken over a spatially diverse range of measurements is fine. Your ears hear those basically like the mic does and we have research that has correlated smoothness of the amplitude response measured on a mic with hearing preferences. The problems just arise from the misuse.

For example when I give my talk on eq or mention it in forums I always say that eq above 500hz should be based on free space measurements over a 180 degree window. Not based on in room nor a single point in space. People bark back that this is preposterous because the in room is what we hear. This is the point I was making earlier. This isn’t true. That single measurement in room isn’t what we hear. In fact even 50 measurements taken in room still isn’t what we hear. It’s a crude approximation.

So I wouldn’t agree that eqing by ear is right and by mic is wrong. I use mics. I just also wouldn’t agree that extreme amounts of eq seeking perfect flatness of measurement is most accurate. It’s very likely that flat in room eq has created a total acoustic mess and our ears hear it as anything but accurate.

 

[highfigh]

I had look that up. https://en.wikipedia.org/wiki/Dunning–Kruger_effect

As I now understand it, everyone suffers from the Dunning-Kruger effect… except me .

Is that the same Kruger as in Kruger Smoothing (Seinfeld reference)?

 

[highfigh]

As great as objective measurement for equalization is, it doesn't guarantee great sound- that's where tweaking by ear comes in, but it has to be done with the knowledge of what works and why.

How many times have we seen comments about systems that looked great on the graphs, but sounded bad? I think more often than not.

Years ago, I was told that the tests for the first seating positions in Audyssey should be repeated, in order for it to make the correct adjustments- this was close to ten years ago.

 

[Johnny2Bad]

Auuuu... I have a slight feeling not many people see you as someone who should post this.

I'm a little stubborn (it's mostly helped me but a few times it's cost me) but I am definitely not married to anything, and there are plenty of things in audio we just don't know enough about so obviously new information is always welcome if it offers some enlightenment.

I know a few other people here that might also be guilty of the same accusation, but we are all learning still.

 

[highfigh]

Measurement mics don’t hear what we hear, but that doesn’t mean they are an invalid tool for scientific or even setup purposes. Most people’s hearing isn’t refined enough (I.e. trained) to use in place of a mic. Even if it was, it would likely be a slower process. That doesn’t mean a person can’t tune a system by ear successfully. The bigger issue is that because people falsely assume that a mic hears what our ears hear they over rely on the measurements. The worst offenders are those who take a single measurement and then eq it flat across the whole band. However it’s not much better to just take a handful of measurements, average then, and do the same. A small amount of eq below 500hz to knock down big peaks taken over a spatially diverse range of measurements is fine. Your ears hear those basically like the mic does and we have research that has correlated smoothness of the amplitude response measured on a mic with hearing preferences. The problems just arise from the misuse.

For example when I give my talk on eq or mention it in forums I always say that eq above 500hz should be based on free space measurements over a 180 degree window. Not based on in room nor a single point in space. People bark back that this is preposterous because the in room is what we hear. This is the point I was making earlier. This isn’t true. That single measurement in room isn’t what we hear. In fact even 50 measurements taken in room still isn’t what we hear. It’s a crude approximation.

So I wouldn’t agree that eqing by ear is right and by mic is wrong. I use mics. I just also wouldn’t agree that extreme amounts of eq seeking perfect flatness of measurement is most accurate. It’s very likely that flat in room eq has created a total acoustic mess and our ears hear it as anything but accurate.

The quandary of using mics- it needs to be calibrated, but how do we know it's REALLY calibrated correctly? It amounts to the act of testing test equipment. Then what? How do we know that the the signal from the mic is up to the abilities of the mic, and what happens after that- how do we reconcile the measured response with what we hear?

Mics and test equipment measure energy, systems like Audyssey, MCACC, YPAO and pro equipment like TC Electronics have circuitry that uses algorithms that are designed to "hear" the room and compensate for it. How does that approximate what happens in our mind when we hear sound? AFAIK, we still don't totally understand all of what's happening.

Measured flat is great in an anechoic chamber, but only to find the speaker's response, which is then tweaked to meet some ideal of what is needed in a real world room, based on trial & error testing. If we're trying to achieve some kind of 'house curve', the in-room response can't be flat, but it can be smooth.

Isn't the goal really to make sure the level of the source material is perceived as equal, assuming the input's level was equal across the frequency band? Our minds are doing a lot of the heavy lifting when we listen- not only does our mind compensate for bad acoustics, it allows us to become accustomed to bad sound and if we're able, we ignore that aspect, to whatever degree we can. That's not something a mic can do.

Ever talk about sound quality with people who are interested in audio systems and have hearing damage from military service, industrial noise or something else? Ask them for opinions on speakers and you might be surprised on their ability to hear many of the same details as those whose hearing is "normal".

 

[everettT]

Anyone remember this?

https://en-de.neumann.com/ku-100

 

[Sheep]

" ... When I discover new facts, I change my opinion. What do you do sir? ..."

-John Kenneth Gailbraith

Ok Dude, you NEED to shrink that signature. It's larger than a pre-masters thesis.

On topic,

Stick it to them Gene.

 

[killdozzer]

Another thought is why does it bother people so much to consider the use of measurement gear? Analyze isn't something I particularly think of as an exercise by ear but rather instrumentation, too....but maybe that's just me. No one was suggesting not to still use one's ears....and many of the responses in that original thread were from people who want to only use their ears exclusively....

That is the first thing that comes to my mind everytime we have one of those "I joined this forum because you all stare at measurement and I use my ears and you should all go out there and listen..." type of BS. You use your ears, honestly!?!? Well whoopty-doo, here we just buy a speaker if graphs are all right and we leave it unpacked and just look at those lovely graphs. That's what we do. Love it or leave it.

 

[killdozzer]

I agree with this but only as half of the story.

As great as objective measurement for equalization is, it doesn't guarantee great sound- that's where tweaking by ear comes in, but it has to be done with the knowledge of what works and why.

Tweaking by ear often means a step away from the great sound towards someone’s personal taste.

How many times have we seen comments about systems that looked great on the graphs, but sounded bad? I think more often than not.

I would be ready to make a bet that you could find more cases where speakers that are perfectly capable of decent sound sound awful because of the entire setup and the owner is ready to jump at you saying that if you think speakers sound bad that’s because your hearing is not as good as his.

Ears adapt. They will always bend their backs under the weight of ones ego. They can adapt to bad sound as well as good. And never ever will you find a guy who’ll say these speakers sound great to me because my ears adapted to bad sound. It’s always the other way around.

The entire Bose sales strategy in having their shops far away from others relies on people’s ears adapting.

Another thing, if you turn on the mic and start registering sound and then mic’s wife comes in starts bitching how he doesn’t register her anymore, but always registers some music and sounds and how he’s not there when she needs him, the mic will NOT start hating those same sounds he enjoyed a second ago.

The fact that his ears are more relevant to one is simply because they are the bottle neck between him and the sound and he has to please them. But that fact tells us very little about the sound and the quality of the speakers. Perhaps he just likes the speakers because they can play as bad as he needs them to play.

The reason I would never rush to please and accommodate my ears is because those same treacherous ears could ask for a completely different set of speakers the very next week.

I try to learn what good speakers are through reading research and then I tell my ears “this is what I want you to adapt to, rather than some Bose, Magnat, kevlar ripping my ear drums, Rega, ZuAudio etc.” As long as you go by ears you might have just adapted to a bad thing like a beginner on an ill tuned guitar playing and still not knowing he is out of tune. And if he is on a desert island he’ll carry on playing out of tune until the world ends.

 

[killdozzer]

Another thing I would like to see debated here is all this "timbre" shite. A D on brass and wood differs because of timbre, your speaker should reproduce both. You should be able to tell them apart. You should be able to tell which is which. Timbre often comes from the body of the instrument, The body of the speaker should be dead. If it's not dead at least one of all those timbres (??? timbres, timbri... sorry, I'll look it up) should be skewd in comparison to the source sound.

 

[shadyJ]

Another thing I would like to see debated here is all this "timbre" shite. A D on brass and wood differs because of timbre, your speaker should reproduce both. You should be able to tell them apart. You should be able to tell which is which. Timbre often comes from the body of the instrument, The body of the speaker should be dead. If it's not dead at least one of all those timbres (??? timbres, timbri... sorry, I'll look it up) should be skewd in comparison to the source sound.

Agreed. Speakers should not have timbre. Any timbre that speakers contribute should be regarded as a flaw.

 

[highfigh]

I agree with this but only as half of the story.

Tweaking by ear often means a step away from the great sound towards someone’s personal taste.

I would be ready to make a bet that you could find more cases where speakers that are perfectly capable of decent sound sound awful because of the entire setup and the owner is ready to jump at you saying that if you think speakers sound bad that’s because your hearing is not as good as his.

Ears adapt. They will always bend their backs under the weight of ones ego. They can adapt to bad sound as well as good. And never ever will you find a guy who’ll say these speakers sound great to me because my ears adapted to bad sound. It’s always the other way around.

The entire Bose sales strategy in having their shops far away from others relies on people’s ears adapting.

Another thing, if you turn on the mic and start registering sound and then mic’s wife comes in starts bitching how he doesn’t register her anymore, but always registers some music and sounds and how he’s not there when she needs him, the mic will NOT start hating those same sounds he enjoyed a second ago.

The fact that his ears are more relevant to one is simply because they are the bottle neck between him and the sound and he has to please them. But that fact tells us very little about the sound and the quality of the speakers. Perhaps he just likes the speakers because they can play as bad as he needs them to play.

The reason I would never rush to please and accommodate my ears is because those same treacherous ears could ask for a completely different set of speakers the very next week.

I try to learn what good speakers are through reading research and then I tell my ears “this is what I want you to adapt to, rather than some Bose, Magnat, kevlar ripping my ear drums, Rega, ZuAudio etc.” As long as you go by ears you might have just adapted to a bad thing like a beginner on an ill tuned guitar playing and still not knowing he is out of tune. And if he is on a desert island he’ll carry on playing out of tune until the world ends.

Ears/hearing change because of exposure to noise/sound, weather and other factors but they don't adapt to the sound differences- that's where the listener's mind steps in and as you wrote, their ego. This is the reason some don't like small or large speakers- because they have preconceptions about small or large speakers. If a small speaker sounds great, it's better to use that than a speaker of a different size, but only if the person can let go of the idea that they're small and that may affect their ego- "bigger is better", right? Personally, I'm far more impressed by a small speaker that digs deep into the bass region than a large one that reaches the same range.

I somewhat disagree about avoiding personal tailoring of the sound for a person's hearing- in order for them to hear the sound 'as intended', they need to do some tweaking because it's well known that everyone's hearing is different, especially if hearing damage enters the scenario. However, that makes the system sound good for one person, not everyone. If someone is the only one to use the system, I think personal EQ is a good thing but as you also wrote, they won't like what they hear in the outside world.

Bose. Yeah. They also require that the dealer not be able to make direct comparisons between their speakers and other brands/models in a store. In reality, I think their Acoustimass system sounds OK for the sound of people speaker, as long as the voice isn't too bassy- then, it becomes very apparent that the satellite cubes and "subwoofer" are in different places. I heard a pair of their outdoor speakers that sounded OK, too. Beyond that, I'm not a fan.

If I have to listen to another mic bitching about her husband's listening habits...........

That said, for systems that move, like the ones for bands and speaking engagements, the system HAS TO be equalized to accommodate for the venue's acoustics- that's not negotiable. Fixed installations (for distributed audio) need to be equalized too, if the desired response is to be achieved. That, too, is for making the changes needed to adjust for the venue because it's not a given that the acoustics will be anything other than OK at best, or bad.

 

[highfigh]

Another thing I would like to see debated here is all this "timbre" shite. A D on brass and wood differs because of timbre, your speaker should reproduce both. You should be able to tell them apart. You should be able to tell which is which. Timbre often comes from the body of the instrument, The body of the speaker should be dead. If it's not dead at least one of all those timbres (??? timbres, timbri... sorry, I'll look it up) should be skewd in comparison to the source sound.

When I bought my first stereo, and I'll be the first to admit that it wasn't very good, I set it up and my dad asked me to play an LP. I put it on and as it played, he had a puzzled look on his face- I asked why and he said that he had always thought the small cymbals were wood blocks because the previous phonographs weren't able to make the instruments sound real.

The whole issue of timbre and overtones is a good argument FOR using a higher sampling rate, IMO.

 

[Matthew J Poes]

The quandary of using mics- it needs to be calibrated, but how do we know it's REALLY calibrated correctly? It amounts to the act of testing test equipment. Then what? How do we know that the the signal from the mic is up to the abilities of the mic, and what happens after that- how do we reconcile the measured response with what we hear?

Mics and test equipment measure energy, systems like Audyssey, MCACC, YPAO and pro equipment like TC Electronics have circuitry that uses algorithms that are designed to "hear" the room and compensate for it. How does that approximate what happens in our mind when we hear sound? AFAIK, we still don't totally understand all of what's happening.

Measured flat is great in an anechoic chamber, but only to find the speaker's response, which is then tweaked to meet some ideal of what is needed in a real world room, based on trial & error testing. If we're trying to achieve some kind of 'house curve', the in-room response can't be flat, but it can be smooth.

Isn't the goal really to make sure the level of the source material is perceived as equal, assuming the input's level was equal across the frequency band? Our minds are doing a lot of the heavy lifting when we listen- not only does our mind compensate for bad acoustics, it allows us to become accustomed to bad sound and if we're able, we ignore that aspect, to whatever degree we can. That's not something a mic can do.

Ever talk about sound quality with people who are interested in audio systems and have hearing damage from military service, industrial noise or something else? Ask them for opinions on speakers and you might be surprised on their ability to hear many of the same details as those whose hearing is "normal".

I think we are basically saying the same things here. I don’t advocate flat, I advocate smooth.

A mics response can actually be measured using fairly rigorous testing methods such that we can be fairly sure that it’s flat. I won’t pretend to fully understand this topic, but here is an article that discusses it.

https://www.bksv.com/media/doc/be1447.pdf
The various methods are all between .01 and .5dB accurate for calibration. I think we do know a microphones response pretty accurately these days, especially at low frequencies.