• 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!

Master Preference Ratings for Loudspeakers

flipflop

Addicted to Fun and Learning
Joined
Feb 22, 2018
Messages
927
Likes
1,240
Bückheim investigated this in some detail. His findings suggest that a weighting factor of more than two would be in order, I think. Do you have AES access? If not I can summarise the findings, if it would help?
I realize that a weighting factor of 2 is a conservative estimate, but I wasn't aware research on the subject had been carried out.
I would very much like a summary of the paper. Unless it's of interest to MZKM, maybe another thread would be more suitable.
 
OP
M

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
4,244
Likes
11,485
Location
Land O’ Lakes, FL
Ah, that is basically how I’d thought it worked. But anyway, you’ve convinced me that AAD would do an adequate job :)

I’d still like to see heavier penalties for peaks than for dips ideally, but can understand if that’s a bridge too far at this stage.
What I got so far:

New Formula 1:
AAD_ON, AAD_LW (normalized), AAD_PIR (normalized), LFX, LFQ
0.93 correlation
pubchart


New Formula 2:
NBD_ON, AAD_ON, AAD_LW (normalized), AAD_PIR (normalized), LFX, LFQ
0.95 correlation
pubchart



So, overall better correlation, I just dislike how the #1 speaker keeps getting a lower score and how the #2 keeps getting a higher score.

Though, looking at the Spins, I would have said #2 is better:

#1:
pubchart


#2:
pubchart


I still have NBD_LW I can add, as well as calculate the normalized NBD for the LW & PIR.

What I would have to do manually is delete individual components to see if some are not well correlated. I would have assumed Sheets could do this (tell me, not automatically), but I guess not.
 
Last edited:

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I realize that a weighting factor of 2 is a conservative estimate, but I wasn't aware research on the subject had been carried out.
I would very much like a summary of the paper. Unless it's of interest to MZKM, maybe another thread would be more suitable.

I'll just put a quick summary here for now, but we can start another thread for it if ppl want to discuss it more :)

Basically, Bücklein added resonances (both peaks and dips) to various signals (music, speech, white noise, and bandlimited speech) and played these back to subjects over headphones. If a subject were able to correctly judge whether or not a stimulus contained an added resonance 10 out of 13 times, this was considered a statistically significant result.

Unfortunately, the results aren't tabulated in a way that permits a general conclusion about the relative audibility of peaks and dips. But there are a few graphs that demonstrate the differences in audibility thresholds at various frequencies and for various Qs. The following results are for music and speech signals:

1606471433124.png


1606471460217.png


And for white noise (Höcker = peak, Senke = dip):

1606471625516.png
 
OP
M

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
4,244
Likes
11,485
Location
Land O’ Lakes, FL
Here is a breakdown of most all current possible components except SM_PIR & NBD_PIR:

chart-2.png

chart-3.png

chart-4.png

chart-5.png

chart-6.png

chart-7.png

chart-9.png
chart-8.png


Let's compare some pairings:

#1 vs #2
#2 beats #1 in every category above except the AAD scores, so those for sure are important.

#6 vs #8
#8 beats #6 in every category (or is very close), except LFX.

#5 vs #13
#13 beats #5 in every category except the AAD scores, LFX & LFQ.

#12 vs #13
#13 beats #12 by a good deal in every category except LFX, but even then it only loses by a bit.

____________________________________________

I still have some other components I can check to see, like say SM on the ON & LW.

Don't want to go too crazy though, as if I'm reading it right, the 13 speakers were ranked from just 18 listeners? Too bad we don't have the data from all 70 speakers.
 

Attachments

  • chart-10.png
    chart-10.png
    37.6 KB · Views: 123
Last edited:
OP
M

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
4,244
Likes
11,485
Location
Land O’ Lakes, FL
Last edited:
OP
M

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
4,244
Likes
11,485
Location
Land O’ Lakes, FL
Here is what I have using the formula stated above using the last 20 measured speakers
Keep in mind the bass issues with the F328Be measurement, it should score higher when re-measured.

Ordered by current formula
Screen Shot 2020-11-28 at 11.55.00 AM.png


Ordered by new formula
Screen Shot 2020-11-28 at 11.55.12 AM.png




@amirm, going by tonal quality only, would you say this is generally better or worse than Olive's formula?


Funny enough, the recently measured R253 got a big boost, mainly due to bass (-10dB: 37Hz); the 8010a is the only high scoring speaker that got punished, mainly due to bass (-10dB: 63Hz).

Here is the CSV with the 7 variables calculated for the 20 speakers.
 
Last edited:

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
Sorry if this has already been discussed (I didn't read this whole thread).

Is it possible to generate what you need to do spinorama and preference rating from the measurements given by Soundstage? If so I wonder if it would be worthwhile to have a community effort to trace the curves and generate the data you need so that these can be included in your database.
 
OP
M

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
4,244
Likes
11,485
Location
Land O’ Lakes, FL
Sorry if this has already been discussed (I didn't read this whole thread).

Is it possible to generate what you need to do spinorama and preference rating from the measurements given by Soundstage? If so I wonder if it would be worthwhile to have a community effort to trace the curves and generate the data you need so that these can be included in your database.
Sadly not. You need 70 points, they only do 6-10.
 

pjug

Major Contributor
Forum Donor
Joined
Feb 2, 2019
Messages
1,775
Likes
1,561
Sadly not. You need 70 points, they only do 6-10.
Thanks. I guess it should have been obvious to me that Soundstage does not have any vertical off axis info.
 

Maiky76

Senior Member
Joined
May 28, 2020
Messages
441
Likes
3,711
Location
French, living in China
For anyone wanting to run their own analysis:
Here is a .csv file with the 11 current variables and whatnot for the 13 bookshelves.

___________
What I have so far:
View attachment 96162

Rating =
24.4294
- 2.21*NBD_ON
- 1.72*NBD_LW
- 0.72*AAD_ON
+ 1.21*AAD_LW
- 0.33*AAD_PIR (normalized)
- 10.76*LFX
+ 0.36*LFQ

Testing
Carmody S2000: 4.9 -> 5.3
Neuman KH 310: 6.2 -> 7.2
Genelec 8341A: 6.8 -> 7.3

Hi,

I probably missed it but how did you finally get the list of "Human score" vs Calculated score?
I had a good look at the data last year I could not figure out for certain which speaker corresponded to a "Human score".
 
Last edited:
OP
M

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
4,244
Likes
11,485
Location
Land O’ Lakes, FL
Hi,

I probably missed it but how did you finally get the list of "Human score" vs Calculated score?
I had a good look at the data last year I could not figure out for certain which speaker corresponded the "Human score".
Dang, I forgot. I think the file napilopez gave simply had them labeled.
 
OP
M

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
4,244
Likes
11,485
Location
Land O’ Lakes, FL
Bass extension (If I recall, the frequency at which the SPL is -6dB relative to the average of the Sound Power).

 
Last edited:

totti1965

Active Member
Joined
Apr 7, 2023
Messages
220
Likes
180
Location
Bonn / Germany
so with / without LFX is with / without subwoofer. I understand! So „Low frequency Extension“ …….why Not simply write „Woofer“?
 

totti1965

Active Member
Joined
Apr 7, 2023
Messages
220
Likes
180
Location
Bonn / Germany
You are doing such a great work! Will you add someday also the speakers which Amir has been tested in recent months, such as the Neumann KH 150?
 
Top Bottom