• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

SPL vs. SQ

OP
Wes

Wes

Major Contributor
Joined
Dec 5, 2019
Messages
1,716
Likes
1,435
Thread Starter #21
Let's pick 2 amps - one has a SINAD of x, the other is only 0.7 x but it is played 0.5 dB louder

which will be preferred?

or use THD levels, etc.
 

Inner Space

Senior Member
Joined
May 18, 2020
Messages
301
Likes
453
#22
For electronics, which will 99 times out of 100 have the same flat FR, the louder component will win.
No question, about electronics. But sometimes I wonder if the same reasoning holds good when confronted by the spectacular trainwreck that is loudspeakers in rooms. When FRs look like mountain ranges, what does level matching even mean? For electronics, we have a pretty good suite of tests and numbers, and we can rely on them. Past the far end of the speaker cables, we don't.
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
10,227
Likes
14,136
#23
I don't think it is JND. JND is a larger difference than .1 db yet .2 db will be enough to make one source sound of better quality.

Now this isn't like an equation where you can do the reverse.

I'll have to find the original work that showed you need .1 db or less difference. I'm thinking it was Vanderkoy and Lipschitz who did it. I've posted it somewhere here on ASR in the past.
 

JohnYang1997

Major Contributor
Technical Expert
Manufacturer
Joined
Dec 28, 2018
Messages
4,785
Likes
7,589
Location
China
#24
No question, about electronics. But sometimes I wonder if the same reasoning holds good when confronted by the spectacular trainwreck that is loudspeakers in rooms. When FRs look like mountain ranges, what does level matching even mean? For electronics, we have a pretty good suite of tests and numbers, and we can rely on them. Past the far end of the speaker cables, we don't.
Average it and match at 500hz. Or use A weighting to calculate an accumulated energy value and match it that way.
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
10,227
Likes
14,136
#25
There is no parameter called "sound quality", but let's simplify it to how loud a speaker can play and its tone, tone just meaning that parts of the FR are different.

I think one of the confusions might be to do with the otherwise straightforward words "quieter" and "louder". These are always frequency-specific when it comes to audio:
View attachment 68490
These are the last three speakers measured by Amir. I got the FR data from @edechamps' Open Loudspeaker Explorer.

Notice that the graph has matched the on-axis FR for all three at 300Hz. So, ostensibly, if you used a 300Hz test tone and matched the SPL, you would get that FR (ignoring the effects of the room). Looking just at 20Hz to 300Hz, what you'll find is that you might like the blue because it's loudest, or because of the emphasis on bass. Or you might like the orange even though it's quieter because that range is flatter and more even. Or you might like red because it reaches deeper and has output at 50Hz stronger than the other two.

So, to summarize, if the tone is exactly the same, assuming it's not pushed beyond what it can handle, the louder speaker will win.

If the tone is different then it's not completely certain what will win.

For electronics, which will 99 times out of 100 have the same flat FR, the louder component will win.
The suggestion is for matching speakers using bandpass limited pink noise 500 hz to 2000hz.
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
10,227
Likes
14,136
#26
Part of a presentation at the AES in the 1980's by Stanley Lipschitz and John Vanderkooy. They showed that under blind listening conditions a level difference of .2 db was detected at mid frequencies in the range of our most sensitive hearing. Previously some suggested .5 db was close enough because of JND's being .7 to 1.2 db. Others thought you should use .25 db just to be sure.

Couldn't find a copy of it right at the moment.
The Great debate: subjective evaluation. from the 1981 AES journal.

For those with AES access it was in this journal.
http://www.aes.org/e-lib/browse.cfm?elib=18966
 

j_j

Addicted to Fun and Learning
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
784
Likes
1,319
Location
My dining room.
#27
It is oft cited on here that a 0.1 dB mismatch in SPL will bias listening tests in favor of the louder device chain.

But the converse does not seem to have been addressed, hence my query
More like .2 dB. But it's not transitive. Slightly more intense almost always gets the nod. Given that slightly more intense signal compared to one that is yet a bit more intense than that, the more intense will get the nod, and so on.

***BUT*** Comparing the 10th step with the original? The ORIGINAL may be considered better.

But I have no idea what SQ means in the OP. If it means "sound quality" then equating "sound quality" to intensity or loudness is like asking which is more orange, an apple or a banana.
 

j_j

Addicted to Fun and Learning
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
784
Likes
1,319
Location
My dining room.
#28
Anyway, to get away from all those complexities just match test tones within 0.1dB electrically at the outputs.
That's always a good idea, but again, JND for tones is different than JND for complex signals, and in fact the result is terribly signal dependent. What's more, the question 'louder or softer' may be completely random, but a preference may exist for (usually) the louder. The problem is complex.
 

j_j

Addicted to Fun and Learning
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
784
Likes
1,319
Location
My dining room.
#29
But surely the OP asks a different question: if an obviously "good" speaker is compared with an obviously "bad" speaker at a precise level match, we can expect (hope?) that the "good" speaker will be preferred. But if the "bad" speaker is inched upward in level, beyond the JND, will it now automatically be preferred? Or is there something about the "good" and "bad" speakers' respective SQs that will allow the original preference to survive the new amplitude mismatch? And if so, what?
You can't possible level match such an experiment. Even if you match levels at one frequency, the two will not have identical frequency response and time characteristics. All you've got left is preference. And yes, by varying the relative levels, you may be able to change preference. Sorry. Now if the bad speaker is REALLY bad, different question, but the first question is then 'why is it bad' (and there are SO MANY ways).
 

j_j

Addicted to Fun and Learning
Audio Luminary
Technical Expert
Joined
Oct 10, 2017
Messages
784
Likes
1,319
Location
My dining room.
#30
The suggestion is for matching speakers using bandpass limited pink noise 500 hz to 2000hz.
Well, that's a good start, but that can go so wrong. The problem (as you know) is that you can't "equalize levels" with two different speakers with different frequency response. All you can do is try to get close. I'd do 100-12K myself, but I don't have any handy reports on that.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
6,598
Likes
12,391
Location
The Neverlands
#31
Let's pick 2 amps - one has a SINAD of x, the other is only 0.7 x but it is played 0.5 dB louder

which will be preferred?

or use THD levels, etc.
The one which is 0.5dB louder. Distortion would have to be really different to pick a preference for it.
SINAD is just an indicator and does NOT describe/have reference to perceived SQ unless noise is audible and or distortion reaches audible levels.

As quite a few have said this is ONLY true for electronics as they do not color the sound (unless they are really, purposely or not) designed to change the sound.
For speakers it doesn't work this way as variations within the audible band are huge compared to what electronics can do.

To answer the OP question in perhaps different wording: Sound quality is not solely determined by level differences. When levels between 0.2 and 0.5dB differences are present it isn't perceived as a difference in loudness when switching between electronics (NOT speakers !) but often as a preference or slight change in 'fullness' of the sound.
That's why, to be certain when comparing electronics one should level match to 0.1dB (1%) to ensure that's not the case.
NOT so with speakers !!!!!
 
Last edited:

pozz

Data Ordinator
Forum Donor
Editor
Joined
May 21, 2019
Messages
2,693
Likes
3,941
#32
You can't possible level match such an experiment. Even if you match levels at one frequency, the two will not have identical frequency response and time characteristics. All you've got left is preference.
Just wanted to highlight this.

Listening to speakers is an imprecise thing, like listening itself, so learning to recognize broad trends in what you like/dislike and having some understanding of how that's linked to the speaker/room combo will help make buying decisions.
 
Last edited:

pozz

Data Ordinator
Forum Donor
Editor
Joined
May 21, 2019
Messages
2,693
Likes
3,941
#33
I don't think it is JND. JND is a larger difference than .1 db yet .2 db will be enough to make one source sound of better quality.
JND is just an experimental parameter in psychology that is defined by results. When testing it will be different from person to person, since it marks when the listener is able to distinguish that something (intensity, duration, pitch, location and so on) has changed. That number produced in the Vanderkooy and Lipschitz study is not the same thing as JND, but related, since JND sets the lower limit of what you can expect while the study established the threshold of preference for slightly louder signals.

You could look at the difference between comparing test tones and comparing entire systems as meaning that the latter has expanded the bandwidth under test and the "net" of differences. The interesting thing here is that during uncontrolled listening comparisons audiophiles fiddle with the volume knob around until they imprecisely set their own "JND", and then pour forth the aesthetic treatise. It could be interesting to set up an entire room for listening sessions using headphone systems and record amateur JND data to get a sense of what the population of audiophiles thinks of as "level matching".
 
OP
Wes

Wes

Major Contributor
Joined
Dec 5, 2019
Messages
1,716
Likes
1,435
Thread Starter #34
The one which is 0.5dB louder. Distortion would have to be really different to pick a preference for it.
SINAD is just an indicator and does NOT describe/have reference to perceived SQ unless noise is audible and or distortion reaches audible levels.

As quite a few have said this is ONLY true for electronics as they do not color the sound (unless they are really, purposely or not) designed to change the sound.
For speakers it doesn't work this way as variations within the audible band are huge compared to what electronics can do.

To answer the OP question in perhaps different wording: Sound quality is not solely determined by level differences. When levels between 0.2 and 0.5dB differences are present it isn't perceived as a difference in loudness when switching between electronics (NOT speakers !) but often as a preference or slight change in 'fullness' of the sound.
That's why, to be certain when comparing electronics one should level match to 0.1dB (1%) to ensure that's not the case.
NOT so with speakers !!!!!

ok, is this an opinion or is there research that supports this?

also, what if one is only 0.2 dB louder?
 

pozz

Data Ordinator
Forum Donor
Editor
Joined
May 21, 2019
Messages
2,693
Likes
3,941

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
6,598
Likes
12,391
Location
The Neverlands
#36
ok, is this an opinion or is there research that supports this?

also, what if one is only 0.2 dB louder?
Aside from research it is my personal observation for which I built a test device about 30 years ago


0.2dB I found a lot harder to pass. 0.1dB I was not able to (using music).
I was only interested in my own audibility capabilities under conditions I was interested in and not having to rely on other research which was hard to find 30 years ago.
Now there are many possibilities to test your own limits.

I built many more test devices, null testers, wobble generator, other AB test devices, peak detectors and meters that could detect very short peaks, output impedance simulators, C1, C2, CE error indicators/counters on my CDP, fake tube amps with very poor quality AB amp secretly in it ... etc.
All have been highly educational and amusing, especially when 'testing' audiophiles without them even knowing nor telling them.

I would advise everyone to find their own hearing limits.
 
Last edited:

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
10,227
Likes
14,136
#37
I think it was @mansr on another forum, a couple years back posted files asking for people to rank them in pairs. As much as is wrong with such online listening tests (variable playback gear and conditions at a minimum), there was one file with near universal agreement in the results. Where two identical files were presented only one was 1 db louder. Not one person commented on a loudness difference, but only on quality differences.
 

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
2,275
Likes
4,606
Location
Hampshire
#38
I think it was @mansr on another forum, a couple years back posted files asking for people to rank them in pairs. As much as is wrong with such online listening tests (variable playback gear and conditions at a minimum), there was one file with near universal agreement in the results. Where two identical files were presented only one was 1 db louder. Not one person commented on a loudness difference, but only on quality differences.
Hmm, I have no recollection of that.
 

Theriverlethe

Senior Member
Joined
Oct 20, 2019
Messages
352
Likes
223
#40
Let's pick 2 amps - one has a SINAD of x, the other is only 0.7 x but it is played 0.5 dB louder

which will be preferred?

or use THD levels, etc.
That's a really complicated question, especially when you throw preference into the mix. SINAD doesn't take psychoacoustics into account. Eg., an amp with a second harmonic at -40dB would get a terrible review from Amir but probably not be noticeable by a human. In that case, the 0.5dB louder amp would be preferred to an amp with 10,000x better SINAD. Assuming that second harmonic reaches an audible level, say -30dB, it might actually be preferred by some people as greater "fullness" or "warmth."
 
Top Bottom