• 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!

Audio Science Review VS Sound & Recording : who has the more accurate on-axis measurements?

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
Amir always points the measurement microphone at the tweeter center, S&R takes into account the manufacturer's specifications (as far as I know), which often means that the mic is aligned halfway between the tweeter and the bass-midrange chassis.
I follow manufacturer spec in the few rare cases where such is specified. Here is an example from Genelec review:
The reference axis was as instructed in the manual: the top of the woofer. I also tested it with tweeter axis and it made no difference.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
If the S&R ground-plane measurements are performed very carefully, they should be superior to those of Amir's NFS in the frequency range above 10kHz, since the conversion of the frequency responses measured in the near field to user defined distances becomes less accurate towards very high frequencies (or needs a lot of data points and time to be accurate up to 20kHz, e.g. 7h measurement for one speaker).
Accuracy error depends on complexity of soundfield, not frequency (not directly anyway). This complexity usually comes about around crossover frequency where multiple radiating sources combine, often with diffraction and other effects. In vast majority of cases though, error is below 1%. Sometimes it rises to 2% and if so, I report on it.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
@amirm at what measurement distance are your measurements calculated? Have you played around with different distances? It would be quite interesting to use the Klippel to illustrate how FR changes with distance, as presumably the data for virtually any distance can be calculated from a single set of measurements - is that correct?
 

MZKM

Major Contributor
Forum Donor
Joined
Dec 1, 2018
Messages
4,240
Likes
11,462
Location
Land O’ Lakes, FL
@amirm at what measurement distance are your measurements calculated? Have you played around with different distances? It would be quite interesting to use the Klippel to illustrate how FR changes with distance, as presumably the data for virtually any distance can be calculated from a single set of measurements - is that correct?
Well, wouldn’t that depend on the room? If you are talking anechoic, wouldn’t all frequencies decrease by 6dB when you double distance?

I think there are programs out there that can simulate in-room performance. Like REW has a multi-sub simulator to show how placement alters the room modes.

What I would love, but what seems not very simple/quick to do, is have the NFS show us the Spinorama with X-degree of toe-in as well as altering seated ear-height. Like, we can get the gist from the H & V off-axis plots, but it would be cool to see the whole story.
 

hardisj

Major Contributor
Reviewer
Joined
Jul 18, 2019
Messages
2,907
Likes
13,908
Location
North Alabama
Simple question:
How does S&R perform their measurements? Anechoic? Ground plane? Quasi anechoic? Some mixture?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,386
Location
Seattle Area
@amirm at what measurement distance are your measurements calculated? Have you played around with different distances? It would be quite interesting to use the Klippel to illustrate how FR changes with distance, as presumably the data for virtually any distance can be calculated from a single set of measurements - is that correct?
That's correct although it is not just one number you modify. It is a large set but I can see if I can modify them all and see what we get. Is there a speaker you want me to test this on?
 

Pio2001

Senior Member
Joined
May 15, 2018
Messages
317
Likes
507
Location
Neuville-sur-Saône, France
What bothers me the most is not the measurements and their differences. It is the attempt to EQ speakers based on measurements whereas other measurements would lead to try other EQ presets.

Hence, the EQ recommendations made here and compiled on this forum can be less pertinent than expected.

Here is what remains when I substract a measurement of my left speaker vs the exact same measurement made with my right speaker.

25 Appariement en.png


We should see two completely flat lines, but in reality, one of my KH-120 is 0.5 dB above the other from 1500 to 3000 Hz, and my two 305P are very different from each other. The stereo image is completely shifted if I swap the left and right speaker, and I should equalize them separately above 1000 Hz.
 

Sancus

Major Contributor
Forum Donor
Joined
Nov 30, 2018
Messages
2,923
Likes
7,616
Location
Canada
We should see two completely flat lines, but in reality, one of my KH-120 is 0.5 dB above the other from 1500 to 3000 Hz, and my two 305P are very different from each other. The stereo image is completely shifted if I swap the left and right speaker, and I should equalize them separately above 1000 Hz.

Not that it's unexpected but that is excellent matching by Neumann and well within their spec for the KH120.
 

hardisj

Major Contributor
Reviewer
Joined
Jul 18, 2019
Messages
2,907
Likes
13,908
Location
North Alabama
That's correct although it is not just one number you modify. It is a large set but I can see if I can modify them all and see what we get. Is there a speaker you want me to test this on?


I’d think a more complex speaker with multiple drive units. A 3-way with low order crossover slopes might be interesting.
 

richard12511

Major Contributor
Forum Donor
Joined
Jan 23, 2020
Messages
4,335
Likes
6,700
Was a pain in the ass to find since I couldn't remember the name of the speaker, but it was the GGNTKT M1.

Ha! I can never remember how to spell that speaker's name. GGTKT? GGTKN? GGKTNT? :p, and it being a fairly new company, google doesn't help me much. Thanks for finding it. I agree with @roland{at}GGNTKT , and I think the problem is even worse for headphones. At least loudspeakers like that have a fairly tight tolerance. Headphones are all over the place, from what I've seen.

Although...that has me wondering if I'd pay 20% more for an individually equalized response :D. I imagine that's gonna give you just about the flattest response ever.

I wonder if Genelec measures every single 8351b? Probably not, but if they do, it would be awesome if GLM could use the monitor's serial number and measurements as a part of its equalization algorithm. That would give GLM a boon that other DRCs (Dirac, Audiolense, etc.) could never have. I did see where @Thomas Lund said that GLM 4 takes into account the speaker's response, but I don't know if he was talking about the speaker as a whole, or each individual speaker. I'm assuming the former, but the latter would be awesome, even if they had to charge a little more for it.

BTW, my new mic came in, so I should be testing the different GLM 4 target curves soon. Tried a couple today, and surprisingly the default curve actually gave me the response that's closest to the actual response of the 8351 in my room, while the 1dB/oct really changed the overall slope. That's all I had time for today(on my lunch break), but I'm excited to test more.
 

Sancus

Major Contributor
Forum Donor
Joined
Nov 30, 2018
Messages
2,923
Likes
7,616
Location
Canada
I wonder if Genelec measures every single 8351b? Probably not, but if they do, it would be awesome if GLM could use the monitor's serial number and measurements as a part of its equalization algorithm.

Well Neumann measures each individual speaker they sell and saves those measurements so it's certainly possible that Genelec does the same. Maybe @Ilkka Rissanen knows?
 

YSC

Major Contributor
Joined
Dec 31, 2019
Messages
3,194
Likes
2,570
Here is what remains when I substract a measurement of my left speaker vs the exact same measurement made with my right speaker.

View attachment 110605

We should see two completely flat lines, but in reality, one of my KH-120 is 0.5 dB above the other from 1500 to 3000 Hz, and my two 305P are very different from each other. The stereo image is completely shifted if I swap the left and right speaker, and I should equalize them separately above 1000 Hz.
I am a bit surprised by the effect on stereo imaging if the shift you mentioned is also true on kh120
 

ROOSKIE

Major Contributor
Joined
Feb 27, 2020
Messages
1,915
Likes
3,394
Location
Minneapolis
What bothers me the most is not the measurements and their differences. It is the attempt to EQ speakers based on measurements whereas other measurements would lead to try other EQ presets.

Hence, the EQ recommendations made here and compiled on this forum can be less pertinent than expected.
Well yes they are not directly pertinant IMHO in terms of a direct port.
They do show that improvements are possible and thus are pertinant in the sense they remind/promote one to use DSP/PEQ ect.
The biggest reveal of all of my in room measurements is that I need to take in room measurements. Those are the responses I EQ, not ones taken elsewhere nor predicted, rather my actual ones.
Some much of the issue of finalizing excellent sound is directly addressing the room anyway that it is impossible not to see the room as a component stereo. It is as integral as the crossovers, drivers, amps and all else.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Well, wouldn’t that depend on the room? If you are talking anechoic, wouldn’t all frequencies decrease by 6dB when you double

Actually, I'm talking anechoically here. As distance from the DUT increases, you move from the nearfield into the farfield. The maths is described in detail here. As a simplified rule of thumb, the "farfield" begins when the mic is around 6 times farther from the speaker than the speaker's largest dimension. IRL, many other factors are also relevant, e.g. the ratio of driver radiating surface areas to baffle dimensions, distances between drivers, crossover slops and frequencies, distances from drivers to baffle edges, etc.

As to whether the inverse square rule applies (6dB/octave decrease in SPL for each doubling of measurement distance), this is also a simplification based on the assumption that the speaker's radiation is 360°. It therefore tends to hold true in the low frequencies, but not quite in the high frequencies (NB: this is a separate issue to the one @dasdoing mentioned related to air humidity and loss).

That's correct although it is not just one number you modify. It is a large set but I can see if I can modify them all and see what we get. Is there a speaker you want me to test this on?

That would be great Amir. I think it would be most interesting to test this with one of the largest speakers you've measured and with one of the smallest you've measured. But no stress ofc :)

Also, are you able to share what the calculated distance is that you normally use in the measurements you publish?
 

dasdoing

Major Contributor
Joined
May 20, 2020
Messages
4,209
Likes
2,674
Location
Salvador-Bahia-Brasil
this is also a simplification based on the assumption that the speaker's radiation is 360°. It therefore tends to hold true in the low frequencies, but not quite in the high frequencies

can you elaborate this?
I can imagine that outside of tweeter height the response will vary depending on distance, but you probably are not talking about this.
if we stay at tweeter height we now have a variation in the relative distance to the other drivers that probably will change interaction of drivers in the crossover region (?)
 

Pio2001

Senior Member
Joined
May 15, 2018
Messages
317
Likes
507
Location
Neuville-sur-Saône, France
I am a bit surprised by the effect on stereo imaging if the shift you mentioned is also true on kh120

I was speaking about the 305P. My setup is asymmetric, with the left speaker next to a bare solid wall, and the right speaker without a wall next to it. Therefore I get more early reflections coming from the left side.
If the loudest JBL is near the wall, the stereo is unbalanced because of too much reflected energy from this side + too loud the speaker. If I swap them, the quietest JBL benefits from the wall's reflections, and the stereo is more balanced.

I didn't notice a problem worth investigating with the KH-120. I just set the amplifier's balance at +0.5 dB to the right side in order to compensate the early reflections from the left side, and it sounds ok.

But the greatest phantom center I heard in my setup was with a complete independent DRC equalization for each channel. The mismatch of the JBL is then perfectly corrected. Too bad I prefer having the direct sound equalized instead of the listening position sound. Equalizing separately the direct sound of both speakers is very time consuming.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
can you elaborate this?
I can imagine that outside of tweeter height the response will vary depending on distance, but you probably are not talking about this.
if we stay at tweeter height we now have a variation in the relative distance to the other drivers that probably will change interaction of drivers in the crossover region (?)

Yeh, exactly. As we move further from the speaker, the relative C2C distances between the tweeter (or whatever point we choose as our reference axis) and the other drivers changes with respect to the mic/listener. Ditto the distance from the acoustic centre of each driver to the edges of the baffle (and indeed, the distance from any point on a driver diaphragm to its surround edge, etc etc).

But as to the specific question regarding the inverse square law, this can best be understood by thinking about the radiation of a sound wave from an omnidirectional point source.

You can visualise a spherical wave emanating from a point source as an expanding sphere, with distance from the source as its radius. Since the area of a sphere = 4πr² (and assuming that acoustical power is equally distributed across the surface of the sphere and that there are no losses), we see that the acoustical power at any point on the surface of the expanding sphere is the inverse square of the distance from the source (a.k.a. the "inverse square law").

The first complicating factor IRL is the one you already mentioned, i.e. that there are losses as the wavefront expands, and that these losses depend on a number of factors (primarily air humidity and frequency).

The second complicating factor is that, except in the case of an omnidirectional source, power is not evenly distributed across the surface of the expanding wavefront. All else equal, the more directional the source, the less power will be lost as the wavefront expands.

Since loudspeaker directivity tends to vary with frequency, as distance from the speaker increases, power loss tends not to be uniform (with, typically, the greatest losses per distance at low frequencies, and the smallest losses at high frequencies).

Having said that, since the non-uniformity of these losses is not usually so large, and since so much is happening in a room with a loudspeaker in reality, this non-uniformity is not often a significant factor, and (for most loudspeaker designs) may be safely neglected once the mic/listener is in the farfield.
 
Top Bottom