• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

The Influence of Listeners’ Mood on Equalization-Based Listening Experience

thewas

Master Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
6,763
Likes
16,237
A 2022 paper from the Acoustics Journal.

Abstract: Using equalization to improve sound listening experience is a well-established topic among the audio society. Finding a general equalization curve is a difficult task because of spectral content influenced by the reproduction system (loudspeakers and room environment) and personal preference diversity. Listeners’ mood is said to be a factor that affects the individual equalization preference. In this study, the effect of a listener’s mood on equalization preference is tried to be investigated. Starting from an experiment with fifty-two listeners, considering five predefined equalization curves and a database of ten music excerpts, the relationship between listeners’ mood and preferred sound equalization has been studied. The main findings of this study showed that the “High-frequency boosting” equalization was the most preferred among participants. However, the “High-frequency boosting” preference of low-aroused people was slightly lower than the high aroused listeners, increasing the preference of the “Low-frequency boosting”.

Source and more: https://www.mdpi.com/2624-599X/4/3/45
 

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,823
Likes
2,951
Location
Ichihara City, Chiba Prefecture, Japan
Thank you for link to interesting research, I will read through the full paper soon.

BTW, even though a little bit out of the scope, we also need to have careful attention on age-dependent hearing decline especially in high-Fq zone. I always implement/keep safe and flexible on-the-fly high-Fq relative gain control for tweeters and super-tweeters adjusting to hearing level of the audience invited to my audio listening sessions as shared in detail in my post here.
 

charleski

Major Contributor
Joined
Dec 15, 2019
Messages
1,098
Likes
2,239
Location
Manchester UK
It's interesting. But I'm always uncomfortable with self-reported scores of mood as these will be inherently subject to variations that are almost impossible to control.
It's a good start, but I'd like to see a test in which, as an example, you set a potentially difficult task for two groups of participants. For one group the task is manipulated behind the scenes so they 'magically' manage to complete it with minimal effort, while the other group is forced to flail around for some time before they suddenly end up completing it without understanding why (fixing a Windows efiboot 0xC0000009a error springs to mind, but that might be going a bit too far ... :mad:). Then you'd be able to be reasonably confident of the actual mood of the participants.
 

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,071
Likes
2,410
Interesting read. The methodology and scientific experimentation transparency is meet regarding this work. No biasing of participants, no aquard and doubtful simulations, independent driver tuning, and proper audiology test included. Only complain I can make is audiology exclusionary condition of - 30 dB on any tested frequency as severe and it would be good to see and study deficiencies direct influence on obtained result individual and on all samples.
So results are quite opposite to Harman or any of so called beloved "curves" for either speakers or headphones.
They are even out of consistency with equal loudness studies done so far in regard of much higher preference for highs and uper highs, higher boost to lows.
Patern is of course small and results are not representative but its a big step in right direction. If anything else it shows how there isn't and can not be a universal preference curve (both objectively and subjectivity influenced) and how previously glorified one's based on worse works (scientifically and methodically) are nothing else than marketing or snake oil if you like.
 

Multicore

Major Contributor
Joined
Dec 6, 2021
Messages
1,677
Likes
1,768
So results are quite opposite to Harman or any of so called beloved "curves" for either speakers or headphones.
They are even out of consistency with equal loudness studies done so far in regard of much higher preference for highs and uper highs, higher boost to lows.
Patern is of course small and results are not representative but its a big step in right direction. If anything else it shows how there isn't and can not be a universal preference curve (both objectively and subjectivity influenced) and how previously glorified one's based on worse works (scientifically and methodology) are nothing else than marketing or snake oil if you like.
I think I read in the last chapter of Floyd Toole's book he recommended the end user make liberal use of such tone controls that may be available.

Do we have somewhere here on ASR a collection of beloved curves?
 

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,071
Likes
2,410
I think I read in the last chapter of Floyd Toole's book he recommended the end user make liberal use of such tone controls that may be available.

Do we have somewhere here on ASR a collection of beloved curves?
Sure you do and it's totally wrong according to this study. Flat for speakers and slapped down (warm) in room and supposed transfer RIF function of same regarding speakers when it comes to headphones, but called (averaged) preference curve regarding headphones even by their own researchers. And preference score in case of speakers. In total disregard to SPL too. Represented as neutral in this paper (and least prefered). As much as I remember his remark whose more tied to specific recordings correction then what it should represent in the end both in the book and hire.
But what's repeatable being presented as horrible, sibling, pricing and awful sounding seams to be only constant preference regarding this work.
 
Last edited:

MAB

Major Contributor
Joined
Nov 15, 2021
Messages
2,138
Likes
4,790
Location
Portland, OR, USA
A 2022 paper from the Acoustics Journal.

Abstract: Using equalization to improve sound listening experience is a well-established topic among the audio society. Finding a general equalization curve is a difficult task because of spectral content influenced by the reproduction system (loudspeakers and room environment) and personal preference diversity. Listeners’ mood is said to be a factor that affects the individual equalization preference. In this study, the effect of a listener’s mood on equalization preference is tried to be investigated. Starting from an experiment with fifty-two listeners, considering five predefined equalization curves and a database of ten music excerpts, the relationship between listeners’ mood and preferred sound equalization has been studied. The main findings of this study showed that the “High-frequency boosting” equalization was the most preferred among participants. However, the “High-frequency boosting” preference of low-aroused people was slightly lower than the high aroused listeners, increasing the preference of the “Low-frequency boosting”.

Source and more: https://www.mdpi.com/2624-599X/4/3/45
Interesting.
The times I think my stereo sounds dull and flat, vs. boosted and bright. Bass sounds tight vs. bloated. I come back another day and it sounds different. I always thought mood is a factor.
Drats! Now I'm gonna be judging my mood every time I think I perceive my hifi's sound has changed!;)
 

Tangband

Major Contributor
Joined
Sep 3, 2019
Messages
2,994
Likes
2,789
Location
Sweden
Interesting.
The times I think my stereo sounds dull and flat, vs. boosted and bright. Bass sounds tight vs. bloated. I come back another day and it sounds different. I always thought mood is a factor.
Drats! Now I'm gonna be judging my mood every time I think I perceive my hifi's sound has changed!;)
Agree. Not only the mood, alcohol or coffe can change the sound , and what we hear as individuals. A slight cold can ruin the perceived sound, making it sound very different. And age HF loss + hearing damage is very common to.

A bit off topic :
I did some interesting ( for me ) comparison with the help of GLM after a ”flat” calibration of my 8340. I shanged the sound manually in GLM mirroring the test frequency results of Kef q150 = a slight peak at 1 kHz and a slight rising treble. With good studiorecordings of female guitar singers , I prefered this. However - with classical music the coloration became tiring in the long run.

 
Last edited:

dualazmak

Major Contributor
Forum Donor
Joined
Feb 29, 2020
Messages
2,823
Likes
2,951
Location
Ichihara City, Chiba Prefecture, Japan

Tangband

Major Contributor
Joined
Sep 3, 2019
Messages
2,994
Likes
2,789
Location
Sweden
A 2022 paper from the Acoustics Journal.

Abstract: Using equalization to improve sound listening experience is a well-established topic among the audio society. Finding a general equalization curve is a difficult task because of spectral content influenced by the reproduction system (loudspeakers and room environment) and personal preference diversity. Listeners’ mood is said to be a factor that affects the individual equalization preference. In this study, the effect of a listener’s mood on equalization preference is tried to be investigated. Starting from an experiment with fifty-two listeners, considering five predefined equalization curves and a database of ten music excerpts, the relationship between listeners’ mood and preferred sound equalization has been studied. The main findings of this study showed that the “High-frequency boosting” equalization was the most preferred among participants. However, the “High-frequency boosting” preference of low-aroused people was slightly lower than the high aroused listeners, increasing the preference of the “Low-frequency boosting”.

Source and more: https://www.mdpi.com/2624-599X/4/3/45
Toole have some interesting comments on equalization :

 

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,071
Likes
2,410
I have hypothesis regarding why we perceive higher levels in treble tiring... on prolonged periods.
It's not because it's bad sounding like that, it's actually opposite and full spectrum sounding. Our brain reacts on what ever is out of the usual spectrum of noise (because it's all only noise to it) with caution and processing it best it can and while it can relately easily ignore low spectrum and least consider mids as noise (vocal area) but is able to easily classifie them as usual and block them. When it comes to highs it's usually a sind of eminent threat (branch braking, a rore a shot...) and it can't discard it no matter how much it wants to (evolution). It's of course tiresome and irritating to it and it does all it can (discomfort, harsh, sibling... and even physical sickness sings) to tell you to stop doing it as it can't cope with it anymore.
This is of course just a hypothesis.
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,938
Likes
3,526
So results are quite opposite to Harman or any of so called beloved "curves" for either speakers or headphones.

You should try a shelving eq setting of +7dB at 200Hz before judging Harman's work. It's a kind of eq that will make people run away screeming.

From the hundreds of forum posts on room eq we know that almost everyone prefers a stronger low boost then the anechoic flat response. And now this study wants us to believe it's actually the other way around?

Something's off here. Maybe our reading of the study is wrong.

Edit 1: Why is the measured frequency respons of the AKG-K52 headphones in the study totally different from what I find elsewhere? If I can believe the other sources their EQ-ing of the headphones is totally wrong, with way to much low end boost. That would explain the preference for the high shelving eq.

Edit 2: Second point regarding the headphone correction: "only frequencies from 100 Hz to 16,500 Hz (frequency range) were inverted with a maximum gain of 8 dB, while frequencies outside this range were damped by 8 dB". So they also implemented a 120Hz low cut (see Figure 2). There's no low end in the sound anymore!

My conclusion is that this study doesn't provide any material to question Harman's preference curves studies. What the study shows is limited to the effect of mood on preference (which was the objective).
 
Last edited:

Multicore

Major Contributor
Joined
Dec 6, 2021
Messages
1,677
Likes
1,768
I have hypothesis regarding why we perceive higher levels in treble tiring... on prolonged periods.
It's not because it's bad sounding like that, it's actually opposite and full spectrum sounding. Our brain reacts on what ever is out of the usual spectrum of noise (because it's all only noise to it) with caution and processing it best it can and while it can relately easily ignore low spectrum and least consider mids as noise (vocal area) but is able to easily classifie them as usual and block them. When it comes to highs it's usually a sind of eminent threat (branch braking, a rore a shot...) and it can't discard it no matter how much it wants to (evolution). It's of course tiresome and irritating to it and it does all it can (discomfort, harsh, sibling... and even physical sickness sings) to tell you to stop doing it as it can't cope with it anymore.
This is of course just a hypothesis.
I have been fascinated by the bandwidth theorem which I learned in college around the same time that I learned the uncertainty principle. I admit to being struck, at the time, by the possibility of mystical truths being related to these. For example I devised a hypothetical relationship between À la recherche du temps perdu-like probability of emotional surprise being inversely related to the information rate of the sensory channel through which the surprising information arrived. All these centuries later I still rather like the idea.

But in a less esoteric way, perhaps the higher information rate of high frequency sounds relative to low frequency sounds is tiresome. A bit like wizz bang TV and movies is more tiresome than looking at a photo album.
 

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,071
Likes
2,410
You should try a shelving eq setting of +7dB at 200Hz before judging Harman's work. It's a kind of eq that will make people run away screeming.

From the hundreds of forum posts on room eq we know that almost everyone prefers a stronger low boost then the anechoic flat response. And now this study wants us to believe it's actually the other way around?

Something's off here. Maybe our reading of the study is wrong.

Edit 1: Why is the measured frequency respons of the AKG-K52 headphones in the study totally different from what I find elsewhere? If I can believe the other sources their EQ-ing of the headphones is totally wrong, with way to much low end boost. That would explain the preference for the high shelving eq.

Edit 2: Second point regarding the headphone correction: "only frequencies from 100 Hz to 16,500 Hz (frequency range) were inverted with a maximum gain of 8 dB, while frequencies outside this range were damped by 8 dB". So they also implemented a 120Hz low cut (see Figure 2). There's no low end in the sound anymore!

My conclusion is that this study doesn't provide any material to question Harman's preference curves studies. What the study shows is limited to the effect of mood on preference (which was the objective).
Filters are based on:
Zhang, D.; Xia, H.; Chua, T.; Maguire, G.A.; Franklin, D.; Huang, D.; Tran, H.; Chen, H. Impact of personalized equalization
curves on music quality in dichotic listening. Digit. Audio Effects-DAFx 2012, 12, 1–7.
and Shen, W.; Chua, T.; Reavis, K.; Xia, H.; Zhang, D.; Maguire, G.A.; Franklin, D.; Liu, V.; Hou, W.; Tran, H. Subjective Evaluation of Personalized Equalization Curves in Music. In Proceedings of the 133rd Convention of the Audio Engineering Society, San
Francisco, CA, USA, 26–29 October 2012.
Apart from midrange one's that are based on ISO 226 2003 and inversion of it.
Will examine used headphone transfer function and hat dependency tomorrow.
Study shows what it shows and I won't be jumping to conclusions.
Edit: seams measurements are fine, this is actual magnitude transfer function not the closest best thing (bass boost).
Magnitude-of-average-HpTFs-for-measured-headphone-types-normalised-to-values-at-1-kHz_Q640.jpg

And you were right regarding understanding and some people.
 
Last edited:

MarkS

Major Contributor
Joined
Apr 3, 2021
Messages
1,062
Likes
1,503
I'm not sure I trust the judgment of "high aroused listeners" ... :p
 

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,938
Likes
3,526
Filters are based on: Z hang...
Doesn't really matter on what principle the filters were based if there's something wrong with the initial response measurement.

Apart from midrange one's that are based on ISO 226 2003 and inversion of it.
And the low end is cut, as explained in the text I quoted and Figure 2 (the black line):
Screenshot_20230302_074355.jpg


seams measurements are fine, this is actual magnitude transfer function not the closest best thing (bass boost).
Don't understand why you share response measurements from other headphones then the ones used in the study. This is the response of the AKG-K52:
Screenshot_20230302_075054.jpg


This is totally different from what you see in their measurements (top graph). And not just a small error, but an opposite response. These headphones already have a strong low end by themselves, and in the study they're pushing that even more.

You should also compare the maximum sensitivity deveation for the headphones. In the study measurements it's about 45dB.That's outrageous. And the variation between left and right. Very unlikely.
 
Last edited:

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,071
Likes
2,410
@Geert it seems unfortunate selection of headphones used that are prone to imperfect seal deviation and most of the time's regarding real use received lose in low end will look like that but not always (for the people who know how to put them on properly which again would be minority).
akg-k52-sellados-minidsp-ears-raw.jpg
HATS used is very precise (more than most) regarding high frequencies measurements but again head model is problematic regarding securing seal which is exactly what happened.
Their newer HATS and discussion:
We don't have in this case possibility to secure headphones ware properly put on with good seal. This could explain deviation from ISO 226 on low levels and we can chose to dismiss the research but it doesn't explain results in regards of preference for higher treble levels especially with lowered down bass levels from imperfect seal.
Much more work needs to be done with as secured good fit (big angular thicker pads) for participants (something no one actually did so far) as possible.
 
Last edited:

Geert

Major Contributor
Joined
Mar 20, 2020
Messages
1,938
Likes
3,526
it seems unfortunate selection of headphones used that are prone to imperfect seal deviation and most of the time's regarding real use received lose in low end

I totally agree, these headphones are known to sit very loose. This might have messed up their initial measurements but also the listening tests themselves. I'm afraid this whole study is seriously influenced by this. Very unfortunate that for a study like this, which certainly involved a lot a work and budget, they could only get their hands on 30€ headphones.
 

ZolaIII

Major Contributor
Joined
Jul 28, 2019
Messages
4,071
Likes
2,410
Methodology is still in infant state in audio researches and not enough foundling to do them on significant (large) scale if you ask me. Hopefully larger community with a lot of work will establish a proper methodology (for which ASR is a good start).
 
Top Bottom