When do measurements/measured differences "not matter"?

[Pareto Pragmatic]

I can certainly think of some examples, but one I saw recently got me thinking.

I saw a comment that, "that narrow of a null won't matter, you won't hear it." And I immediately agreed. But how narrow is narrow? So how narrow of a null is narrow enough "not to matter"? Is there a general rule? I don't know.

But I do know that comments that some measurements that don't matter, or are nothing to worry about, show up regularly. So it might be useful to have a thread that puts those ideas in one place.

So when is a measurement something one can ignore safely?

 

[Purité Audio]

It’s not the measurement but the result, regarding your question it is difficult to determine the thresholds of audibility.
Keith

 

[boxerfan88]

I saw a comment that, "that narrow of a null won't matter, you won't hear it." And I immediately agreed. But how narrow is narrow? So how narrow of a null is narrow enough "not to matter"? Is there a general rule? I don't know.

But I do know that comments that some measurements that don't matter, or are nothing to worry about, show up regularly. So it might be useful to have a thread that puts those ideas in one place.

So when is a measurement something one can ignore safely?

I’ll share my perspective:
* if null doesn’t show up in MMM measurement most likely won’t be heard
* if sweep measured about 4”-5” apart (head width) shows the null shifted, most likely our brain can fill it in

 

[Jim Taylor]

I can certainly think of some examples, but one I saw recently got me thinking.

I saw a comment that, "that narrow of a null won't matter, you won't hear it." And I immediately agreed. But how narrow is narrow? So how narrow of a null is narrow enough "not to matter"? Is there a general rule? I don't know.

But I do know that comments that some measurements that don't matter, or are nothing to worry about, show up regularly. So it might be useful to have a thread that puts those ideas in one place.

So when is a measurement something one can ignore safely?

I have often used the phrase, " ... below the threshold of audibility." For me, that's the criterion I use for safely ignoring a measurement.

These articles examine the problem from the perspective of phase. They might give you an idea of how complicated the problem of audibility is:

RealTraps - Audibility of Narrow-Band EQ

realtraps.com

Testing Loudspeakers: Which Measurements Matter, Part 2

Joe D’Appolito shares his experiments with loudspeaker evaluation.

audioxpress.com

https://pearl-hifi.com/06_Lit_Archive/14_Books_Tech_Papers/Bauer_Benjamin/Audibility%20of%20Phase%20Distortiona.pdf

I would note that whenever we listen to the output of a speaker in a room, we are listening to nulls at various frequencies caused by room reflections. A listener should be able to move around the room and test themselves regarding the audibility of nulls ... especially ceiling bounce ... rather easily, using both test tones and music.

You might be surprised.

 

[Keith_W]

I don't want to sound dismissive, but you are asking a massive question about audibility thresholds. There are entire books written on the subject of what we can perceive and what we can't, and the hows and whys. What you need is a psychoacoustics textbook like Moore. Dr. Toole's book probably has enough info on psychoacoustics to satisfy a beginner, but my criticism of his book is that he does not go into enough detail about some of the debates. He puts his own spin on it, then he comes to his own conclusion. If you want to delve into it, you have to read the studies that he cites.

 

[Old_School_Brad]

If you’re curious about nulls and peaks, you can easily test them on your system by using a high-Q EQ and applying boosts or cuts of +10 to +15 dB at various frequencies.

 

[Pareto Pragmatic]

I don't want to sound dismissive, but you are asking a massive question about audibility thresholds.

You are of course correct, and I have delved into reading about these issues now and again. Mostly here, but occasionally looking at links like the one you and Jim Taylor provided (and thanks for those).

Most of what I have seen has focused on thinks like noise and distortion, and how "-xx dB below signal is inaudible". Which is fine, useful information.

My goal here is not to invite complete explanations, but rather to see what people who have such knowledge would see something and immediately dismiss it or see it as a concern. I do see this a lot in threads, btw. "Don't worry about that bit, focus on these other bits".

As a practical example, I am limited to 10 PEQ settings with my Wiim. I, of course, need 12 or 13 to do the job completely. So I have to prioritize, in a situation where I have room issues that go up to 2k (don't ask, it's complicated.) Knowing that one "problem" is small or not a real problem would help.

So things like this:

* if sweep measured about 4”-5” apart (head width) shows the null shifted, most likely our brain can fill it in

I particularly like this example since it connects measures to the experience of listening.

So let me offer this. Given a null, it might cover 1/8th of an octave, or 1/32nd of an octave. I would eq 1/8th octave, but I see no reason to eq a dip that covers 1/32nd of an octave no matter how bad it is. Am I wrong? And how narrow is narrow enough to ignore?

 

[Old_School_Brad]

You are of course correct, and I have delved into reading about these issues now and again. Mostly here, but occasionally looking at links like the one you and Jim Taylor provided (and thanks for those).

Most of what I have seen has focused on thinks like noise and distortion, and how "-xx dB below signal is inaudible". Which is fine, useful information.

My goal here is not to invite complete explanations, but rather to see what people who have such knowledge would see something and immediately dismiss it or see it as a concern. I do see this a lot in threads, btw. "Don't worry about that bit, focus on these other bits".

As a practical example, I am limited to 10 PEQ settings with my Wiim. I, of course, need 12 or 13 to do the job completely. So I have to prioritize, in a situation where I have room issues that go up to 2k (don't ask, it's complicated.) Knowing that one "problem" is small or not a real problem would help.

So things like this:



I particularly like this example since it connects measures to the experience of listening.

So let me offer this. Given a null, it might cover 1/8th of an octave, or 1/32nd of an octave. I would eq 1/8th octave, but I see no reason to eq a dip that covers 1/32nd of an octave no matter how bad it is. Am I wrong? And how narrow is narrow enough to ignore?

Why not compare the different settings and see if you can hear a difference?

 

[sergeauckland]

You are of course correct, and I have delved into reading about these issues now and again. Mostly here, but occasionally looking at links like the one you and Jim Taylor provided (and thanks for those).

Most of what I have seen has focused on thinks like noise and distortion, and how "-xx dB below signal is inaudible". Which is fine, useful information.

My goal here is not to invite complete explanations, but rather to see what people who have such knowledge would see something and immediately dismiss it or see it as a concern. I do see this a lot in threads, btw. "Don't worry about that bit, focus on these other bits".

As a practical example, I am limited to 10 PEQ settings with my Wiim. I, of course, need 12 or 13 to do the job completely. So I have to prioritize, in a situation where I have room issues that go up to 2k (don't ask, it's complicated.) Knowing that one "problem" is small or not a real problem would help.

So things like this:



I particularly like this example since it connects measures to the experience of listening.

So let me offer this. Given a null, it might cover 1/8th of an octave, or 1/32nd of an octave. I would eq 1/8th octave, but I see no reason to eq a dip that covers 1/32nd of an octave no matter how bad it is. Am I wrong? And how narrow is narrow enough to ignore?

My understanding based on the 'wisdom' that was current thinking when I trained in the early 1970s, was that our hearing is essentially 1/3rd octave, which is why equalisers in those days were called third octave equalisers and had 31 frequencies. I have always used third octave smoothing on loudspeaker frequency response curves, as that was my understanding of what they would sound like. One sixth octave smoothing would show up anything that might intrude but unlikely to be audible. Any smoothing less than that was essentially for manufacturers to see if there were any unwanted effects. This was before waterfall plots showing energy storage were available.

S.

 

[Pareto Pragmatic]

Why not compare the different settings and see if you can hear a difference?

As a practical matter, I can see this, combined with this:

If you’re curious about nulls and peaks, you can easily test them on your system by using a high-Q EQ and applying boosts or cuts of +10 to +15 dB at various frequencies.

... and using headphones to compare audible issues.... yes that would work. But it's more work than looking at a sweep and going from there.

My understanding based on the 'wisdom' that was current thinking when I trained in the early 1970s, was that our hearing is essentially 1/3rd octave, which is why equalisers in those days were called third octave equalisers and had 31 frequencies. I have always used third octave smoothing on loudspeaker frequency response curves, as that was my understanding of what they would sound like. One sixth octave smoothing would show up anything that might intrude but unlikely to be audible. Any smoothing less than that was essentially for manufacturers to see if there were any unwanted effects. This was before waterfall plots showing energy storage were available.

Interesting, and thanks for that. It makes me feel better about mostly using 1/6 smoothing for developing my EQs.

Waterfall/spectrograph information is definitely essential for me, btw. And I go narrower on certain things based on those measures for sure.

 

[Keith_W]

So let me offer this. Given a null, it might cover 1/8th of an octave, or 1/32nd of an octave. I would eq 1/8th octave, but I see no reason to eq a dip that covers 1/32nd of an octave no matter how bad it is. Am I wrong? And how narrow is narrow enough to ignore?

The answer depends on how deep and how wide your null is, and where on the frequency band it occurs. Read Equivalent Rectangular Bandwidth or watch this video.

 

[tmuikku]

Hi,
to get some reference what is relatively more or less audible, try https://harmanhowtolisten.blogspot.com/2011/01/welcome-to-how-to-listen.html listening skill test software.
I bet you and everyone else scores far better on EQ peak test than in EQ dip test, the dips can be very tough to notice.

Now, what ever is your top score on the EQ test after practice and trying to push it, you'd be sure that if your speaker has less of a dip, narrower band and less deep, you'd very likely not notice it. If it was wider and deeper, you'd likely notice it.

Make sure to similarly train on the EQ peaks, you'll learn they are relatively easy to notice and quite annoying, so in general any peaks in your speaker response are likely audible and very much worth weeding out, and any narrow band dips are relatively inaudible and very likely not an issue.

ps. there was info somewhere that level 8 or 9 at any test would make one qualify into "trained listener" group on that kind of tests in Harman listening related studies. This is also useful information to get some kind of a reference on listening skill with various aspects of your system. For example, if you think your system LR balance is off, and you test full score on pan test, then you probably are correct. If you do not score very well on that test, you probably should not trust what you hear but perhaps experiment a little to make sure it's not the imagination playing tricks. Have fun with it!

 

[Pareto Pragmatic]

Hi,
to get some reference what is relatively more or less audible, try https://harmanhowtolisten.blogspot.com/2011/01/welcome-to-how-to-listen.html listening skill test software.
I bet you and everyone else scores far better on EQ peak test than in EQ dip test, the dips can be very tough to notice.

Now, what ever is your top score on the EQ test after practice and trying to push it, you'd be sure that if your speaker has less of a dip, narrower band and less deep, you'd very likely not notice it. If it was wider and deeper, you'd likely notice it

Personally, I have a much easier time hearing differences with dips in terms of the effects of other frequencies around it. A narrow null at 60 I will first hear in terms of more clarity in vocals, and it takes a LOT of concentration to hear the issue directly, even with a "walking bass" line playing.

I am not sure I want to train my listening more than I have. I can't unhear a problem once heard, and do I want to go down that path? A path that leads with certainty to miniDSP or something like it? Maybe. But for now, I like what I have going on, so why make trouble for myself?

 

[tmuikku]

Ignorance is bliss

I think that If you want to make your system better there really is no other way than to improve listening skill. Basically, listening skill sets limit what you know is good enough, and any differences beyond reliable listening skill is just random. Only way to sort whether something is for better of for worse, is to improve listening skill and figure out if there truly is a difference doing AB tests. If you don't have any understanding, or touch, what the level of listening skill is, it's impossible to be deterministic with what you hear is true or not so it's all just random. If this is what you want it's all good, it's a hobby after all

it's everyones own decision what you wanna from the hobby. If one cannot reliably listen, then it's the gear game roll stuff in and out, looking for something but doing it blindfolded. You are being sold all kinds of stuff in the name of better sound because you cannot self determine whats better, good for business and bad for your wallet. Although, money is used on all sort of silly things every day so, this is fine as well.

Power is to be awake on it, so that you are doing what you prefer, and not what the marketing department prefers essentially exploiting you, consumers consuming endlessly. Just the realization of this shifts power back to you.