• 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.

Should we (I) get into speaker testing & measurement

Should we get into proper speaker measurements?

  • Yes

    Votes: 147 74.2%
  • Yes, but do it later.

    Votes: 19 9.6%
  • No. Stay with Electronics.

    Votes: 32 16.2%

  • Total voters
    198

SIY

Major Contributor
Technical Expert
Joined
Apr 6, 2018
Messages
2,714
Likes
4,634
Location
Phoenix, AZ
It was frustrating to me that no matter what I did to position the headphones, the response still never looked right. If a measurement is THAT fussy and will have that much operator dependence, its gauge R&R will be laughable. As in "absolutely unreliable."

In any of my measurements, I averaged several removal/repositioning, and in fact AP wrote a macro for that very purpose.

The bright side is that at least for comparative measurements within a single fixture/operator, one can discern relative differences.
 

Mad_Economist

Active Member
Joined
Nov 29, 2017
Messages
103
Likes
141
It was frustrating to me that no matter what I did to position the headphones, the response still never looked right. If a measurement is THAT fussy and will have that much operator dependence, its gauge R&R will be laughable. As in "absolutely unreliable."

In any of my measurements, I averaged several removal/repositioning, and in fact AP wrote a macro for that very purpose.

The bright side is that at least for comparative measurements within a single fixture/operator, one can discern relative differences.
I must admit, I've never worked with the AECM206 before - it's possible that some element of its design allows additional leaks which are greater than is typical of the GRAS 45CA it's patterned after. For the systems I have worked with, I've achieved variation in response across measurement sessions which was at or below 1dB at lower frequencies using the same headphones (and a fairly wide variety of said), which I would consider acceptable.

Repositioning in my opinion is mostly an attempted fix for high-frequency response variations - if you start from a poorly coupled position, moving into a well coupled position by accident is pretty rare IME (assuming, as I have been this whole time, that your issue is with the low frequencies).

If you don't trust it on the basis of low-frequency behavior, unfortunately there is no safety in comparative measurements, as this property is highly variable across headphones depending on their acoustics. An older Hifiman design can have an air gap of 5mm between the pad and the head and still measure relatively flat down to 30-40hz, whereas an AKG K553 will roll off from 300hz with a slight leak and stay linear to below 40hz when well coupled.
 
Joined
Mar 27, 2019
Messages
81
Likes
94
So my question is do we have enough people on this site to fund this? How much do I need to donate? It sounds like Amir needs $70K.
Speakers=yes
It’s 100K. Ie. Not 10K and not a $1M.
It’s about how much a typical small app would cost in development.

It might sound a lot of capital for a single person on a salary, but not a company. I’m sure Amir could swing it. I could swing it. It’s peanuts compared to say... starting your own speaker company...

The question is- how much interest/attention would there be in this enterprise?

A vote for a monetary donation is really a vote of genuine interest. As in, are you willing to put money where your mouth is?

But if attention is a proxy for money, then I say with 10 pages of responses in just 24 hours, the answer is a resounding yes!

But what is the mission of ASR?

Are we non-profit? Or not for profit? Are we an educational organisation seeking to advance the science of sound?

Or are we just building up membership, getting attention, then monetising?
Eg. Click-bait, Advertising, Subscription

if it’s the first, Count me in.

If it’s the second. In the way some “free” services or communities have become, I’ll pass.
 
Last edited:

Matias

Member
Patreon Donor
Joined
Jan 1, 2019
Messages
96
Likes
65
Location
São Paulo - Brazil
In my opinion Amir should build a room similar to Dynaudio's new Jupiter room: 13 meters in all directions, 360° measurement with 31 mics mounted on a robot arm. It's like "go big or go home". :D

https://www.dynaudio.com/dynaudio-a...jupiter-delivers-dynaudio-s-trademark-quality




Just kidding. I also voted "No" because of all the issues involved. Better to focus and expand the electronics database.
But headphones would be a lot more realistic, specially if done with Harman's standards.
 
Joined
Mar 27, 2019
Messages
81
Likes
94
In my opinion Amir should build a room similar to Dynaudio's new Jupiter room: 13 meters in all directions, 360° measurement with 31 mics mounted on a robot arm. It's like "go big or go home". :D

Just kidding. I also voted "No" because of all the issues involved. Better to focus and expand the electronics database.
But headphones would be a lot more realistic, specially if done with Harman's standards.
I think some may have missed the evolution of what the Klippel system is able to do. To take speaker measurements traditionally required a large anechoic chamber (cost ~$1M) OR a large outdoor field, with the speaker perched on a stand as far away from the ground as practically possible, with the nearest allowable object some ~10m away, simultaneously measuring during near silent atmospheric conditions OR a measurement taken indoors at 2 different mic locations and then deliberate and careful blending of the bass/mid-bass responses to the midrange/treble by a manual operator.

The Klippel NFS can do all of this in a safe small environment like an office/bedroom sized space in an automated / batch style fashion.
At $70K, it’s a bargain.

Remember that a typical measurement for a full size speakers takes an hour just to setup for scenarios 2 and 3 aforementioned. Sure, if you have a million dollar anechoic chamber, it’s much faster to setup and go.

My bet is the NFS is so efficient that Amir’s time to unpack the speakers, lift it up and place it onto desired point, and hitting Start, get out of the way to let it do it’s job, then lift it off, take it down, and pack it up and print the shipping label will now be as significant a time constraint as the measurements process itself.

This Klippel hardware/software based solution will automagically churn out full frequency response measurements in complete CEA2034/Harman spinorama fashion faster than he can get speakers in and out!

Amir will be able to do in minutes what others without an anechoic chamber have done in hours!!

(Some background reading of taking speaker measurements indoor)
http://audio.claub.net/software/FRD_Blender/White Paper - Accurate In-Room Frequency Response to 10Hz.pdf
 
Last edited:

DDF

Active Member
Joined
Dec 31, 2018
Messages
239
Likes
227
If you don't trust it on the basis of low-frequency behavior, unfortunately there is no safety in comparative measurements, as this property is highly variable across headphones depending on their acoustics. An older Hifiman design can have an air gap of 5mm between the pad and the head and still measure relatively flat down to 30-40hz, whereas an AKG K553 will roll off from 300hz with a slight leak and stay linear to below 40hz when well coupled.
Many years ago I was professionally involved with headphone design: setting repeatable test method standards, executing testing on Hats, kemar, various telecom pinnae, testing human subject hrtf with etymotic in ear mics in anechoic chambers, working with headphone suppliers to execute custom designs etc.

IME, reducing the variability in the low frequencies caused by variance in seal requires using a headphone with low acoustic output impedance. Its a very viable design choice though it does come with some trade offs such as typically poorer isolation. If reviewing online headphone test results, designs showing less low frequency isolation may also provide more consistent low frequency spectral balance with repeated placement
 
Joined
Mar 18, 2019
Messages
39
Likes
39
First of all: I would love to see some good and detailed speaker measurements!

A lot of speaker measurements you see, if you find some, are not very detailed. So in my opinion you can't really rely on those.
If ASR should get into speaker measurement, the measurements have to be as detailed as possible, they need to be repeatable and reliable. Otherwise they could not keep up to the scientific approach we all love about this site.
So the Klippel system @amirm posted seems to be a really good solution.

The following measurement paramters come to my mind when thinking about speaker measurements, wich have been mentioned before:
- frequency response all around the speaker
- distortion
- dynamic compression
And after looking at Klippels website there is a lot more that can be measured.

But there are some problems I see and also others already figured out in this thread:
- shipping
- getting the funds
(- rating the speakers as it is done with the SINAD)
-> Tooles preference rating (but I think it is not as definitive as the SINAD)
- showing and writing down the results, because you get a lot of data from frequency response all around the speaker alone
-> maybe limit the review to some specific points, but making the complete data accessable?
- amirms time

So the funds aside it mostly depends on @amirm. So if you want to do it @amirm, than do it. But you should really think about it again, wich I guess you will do not only once.

Niklas
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
20,088
Likes
23,264
Location
Seattle Area
But there are some problems I see and also others already figured out in this thread:
- shipping
While I have access to a few speakers and can draw on local resources for more, let's have a show of hands of who is willing to send in one speaker for testing. Yes, I only need one, not a pair.
 

DDF

Active Member
Joined
Dec 31, 2018
Messages
239
Likes
227
While I have access to a few speakers and can draw on local resources for more, let's have a show of hands of who is willing to send in one speaker for testing. Yes, I only need one, not a pair.
It would be great to spot test a second speaker from a pair. Testing only 1 would miss one of the most important attributes of a speaker system for imaging, which is pair matching. Driver variance can be significant for drivers taken from different manufacturing batches
 

Mad_Economist

Active Member
Joined
Nov 29, 2017
Messages
103
Likes
141
Many years ago I was professionally involved with headphone design: setting repeatable test method standards, executing testing on Hats, kemar, various telecom pinnae, testing human subject hrtf with etymotic in ear mics in anechoic chambers, working with headphone suppliers to execute custom designs etc.

IME, reducing the variability in the low frequencies caused by variance in seal requires using a headphone with low acoustic output impedance. Its a very viable design choice though it does come with some trade offs such as typically poorer isolation. If reviewing online headphone test results, designs showing less low frequency isolation may also provide more consistent low frequency spectral balance with repeated placement
Absolutely agreed, although I reckon that directly including a metric in testing for variation in response with added leaks or positional variations is both a bit more consistent as a predictor and more intuitive to users.
 

Matias

Member
Patreon Donor
Joined
Jan 1, 2019
Messages
96
Likes
65
Location
São Paulo - Brazil
While I have access to a few speakers and can draw on local resources for more, let's have a show of hands of who is willing to send in one speaker for testing. Yes, I only need one, not a pair.
Ideally you would like to test both for manufacturing consistency between pairs, which is important for channel matching.
 

DDF

Active Member
Joined
Dec 31, 2018
Messages
239
Likes
227
Absolutely agreed, although I reckon that directly including a metric in testing for variation in response with added leaks or positional variations is both a bit more consistent as a predictor and more intuitive to users.
I agree. I was just hoping to provide some useful advice to purchasers, in the absence of testing.

For objective testing, we used to use a calibrated "leaky coupler" which added a known and calbrated amount of leakage to assess sensitivity to leakage loss. Alternatively, we also used to use calibrated pressure for pushing the headphone against the artificial ear, but back then, all artificial ears were far too stiff to get an accurate assessment of real-use leakage. That's one place where the in ear mic and real user came in. I once ran a study similar to Dr Olives investigating leakage on real heads, but it was far more limited in scope both in terms of headsets investigated, and subjects used as guinea pigs. Results were in line with Dr. Olive's.
 

BoredErica

New Member
Joined
Jan 15, 2019
Messages
1
Likes
0
My largely uneducated opinion is testing speakers is the best idea if it can be pulled off. Transparent amp/dacs aren't that hard to make. The difference from one amp/dac to another shouldn't be much. But with speakers I bet the difference will be larger and I find it confusing which speakers to get. I personally don't see much point in reviewing a bunch of transparent gear when there are more relevant gear choices when it comes to audio as a whole.

I don't even know where to go for speaker reviews without the woo woo.

If a good setup is done for measuring speakers that most people agree is good, then I'd be willing to submit my speakers for testing. But... I just have some random Logitechs and a Krk 6 G2. Not very desirable speakers.
 
Last edited:

D700

Active Member
Patreon Donor
Joined
Dec 21, 2018
Messages
123
Likes
85
@amirm I voted yes but later. I would challenge you however to only pursue it if you think you could objectively show the differences between say for example, an LS50, a Magnepan 1.7i and a Klipsch RF-7iii...and why one might be more suited for a certain room or style of music. Because measuring frequency range at a given avg SPL for those 3 don’t tell much of a story...especially in the Dirac era approaching. I’d like to hear how you differentiate those 3 with objective tests before contributing $$$. Maybe mock-up a comparison test between 2 hypothetical speakers so we can see it?

Not to mention, how do you measure impacts of room size, placement with a given design?
 
Last edited:

Frank Dernie

Major Contributor
Patreon Donor
Joined
Mar 24, 2016
Messages
1,588
Likes
2,007
Location
Oxfordshire
Assuming so, would people want me to go after headphone testing first???
Personally no. I still think the variation in our pinna shape effects whether 'phones sound good and the volume enclosed by the earcup is so tiny coming up with any compensation for this seems unlikely.
The correction made by my AKG Q90s for me is marked, and it is just for me, not anybody else. I don't believe any 'phones can be good for everybody so I think you would be just eliminating rubbish by a general test.
 

jhaider

Senior Member
Joined
Jun 5, 2016
Messages
303
Likes
182
I would love to see a lot of speakers tested using a Klippel NFS. However, there are some practical considerations. The most important is the high probability of transit damage, even for speakers in original packaging. One example: I sold a KEF system with 5 eggs and a subwoofer on eBay and shipped over FedEx. I took impedance sweeps on all speakers before I shipped. They shipped in the original KEF box, cloth wrappings, and styrofoam clamshells. The buyer asked for a return because one didn't work. I agreed, and when I received them I noticed that the crossover network on one egg was cracked and one of the inductors had fallen off. Others had inductors loose as well, and also longer worked. The sub also powered on, but shut off shortly after the music started. That must have been some drop! I have other example but that was the most egregious.

For reference, here's how I currently measure speakers on our freestanding deck, which is also the roof of our 2-car carport. My limit is about 40lbs due to the camera ball and my willingness to hoist something heavy 6' off the ground. We live in a city neighborhood, so lots are small - the price to be able to walk to great restaurants, our daughter's piano lessons, several parks, etc. - and I confirm that my neighbors are out before measuring. The other downside to measuring speakers is that it's loud! (And yes, that poor deck has been properly washed and stained since I snapped this picture!)

IMG_0514.jpeg

IMG_0515.jpeg
 
Joined
Mar 5, 2018
Messages
32
Likes
17
Location
Germany
yes of course. I like to see some measurements from active monitors. for example jbl 305, adam t7v, yamaha hs7 and kali audio lp-6.
 

617

Active Member
Joined
Mar 18, 2019
Messages
137
Likes
119
Location
Somerville, MA
It would be great to spot test a second speaker from a pair. Testing only 1 would miss one of the most important attributes of a speaker system for imaging, which is pair matching. Driver variance can be significant for drivers taken from different manufacturing batches
Pair matching is not a trivial issue in inexpensive speakers. My 305 mkii which I measured here have immense response variation.
 
Top Bottom