• 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!

Excellent Presentation on Headphone Measurements

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,590
Likes
239,537
Location
Seattle Area
Very well done presentation from our industry member @Dan Clark on headphone measurements. It is done at just the right level of detail and accuracy so I highly suggest watching it:

 

Eldus

Active Member
Joined
Jun 8, 2021
Messages
132
Likes
114
That was very interesting.
 

robpbg

Member
Joined
Mar 26, 2021
Messages
18
Likes
21
Good stuff. More focus on issues beyond frequency response is needed, as per his slide in the video. Distortion measured as a % THD (vs db below which is harder, for me, to appreciate) at various levels including peaks that correspond to 90db at a 10' or so listening distance would be easy (Erin does this at times like on his f226be review). More analysis of impulse response / decay time / ringing. etc per Dan's bullet list. In short, measure what's important not just what's easy to measure.
 

ExUnoPlura

Member
Forum Donor
Joined
Jul 23, 2020
Messages
54
Likes
139
Location
Oregon Coast
Indeed, the open issues segment was enlightening, including the audience comment that VR/AR and spatial audio concepts may drive further research. NIST could be leading the charge if only there was a national security or competitive aspect…
 
  • Like
Reactions: ADU

robpbg

Member
Joined
Mar 26, 2021
Messages
18
Likes
21
There is a video (that I can't find again unfortunately) done by a prof at Yale (edu, not some bs company) that highlighted impulse response... turns out to be very important. Hopefully someone can find the vid and link to it.
 

kongwee

Major Contributor
Joined
Jan 22, 2022
Messages
1,024
Likes
276
At 26:37. Soundstage. imaging. We can heard definitely we can measure. Just nobody will touch what kind of measure unit, it is good or bad.
 

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
Dan is obviously a very smart guy. But this was a little more advanced than 101. And he glossed over alot of important details in his desire to cover alot of ground over a relatively short space of time. I felt like I gained alot more actual knowledge about measurements and how to begin to read and interpret them from Sean Olive and Tyll Hertsen's seminars. And certainly Dr. Toole's presentation on speakers.

One of the problems with having a headphone developer like Dan do a presentation like this is that they can't be quite as forthcoming with all of the info that they know and are using in the development of their products for reasons of propriety. He touches on a number of topics du jour though in this. And I'd encourage people who are really interested in these subjects to delve more deeply into them on their own, or here on the forum. And to do some more research to get a more full and complete picture of some of the things he's talking about.

I appreciate the link though on this. Because I still haven't made it to any Canjam's yet myself. And it's always sort of interesting to hear Dan's take on things.

A livestream with Dan would be interesting some time. Maybe ASR, or the Headphone Show could set something like that up.
 
Last edited:
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,590
Likes
239,537
Location
Seattle Area
Distortion measured as a % THD (vs db below which is harder, for me, to appreciate) at various levels including peaks that correspond to 90db at a 10' or so listening distance would be easy (Erin does this at times like on his f226be review).
I don't know who you are referring to but I already show headphone distortion in percentage:

index.php


And it is shown at three different levels so you can see the progression. I don't know what you mean by distance though. These are headphones so distance is constant.
 

ReluctantGT

Member
Joined
Nov 9, 2021
Messages
16
Likes
37
Thanks for sharing! The most interesting thing I've watched in a while. His remarks on distortion in particular. I thought he did a great job not pimping his products too hard. You can tell how proud he is of the Stealth, though.
 

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
You can tell how proud he is of the Stealth, though.

Indeed! His presentation was a bit tilted towards the DCA philosophy of headphone development. But that's to be expected from any developer.

The presentations by Sean Olive and Floyd Toole are also a tad skewed in that way, imho. Dr. Toole made the publication of some of their research one of the conditions of his employment at Harman though, which is a bit different. He shows a similar kind of pride though in some of Harman's products (like the JBL M2) in his presentation on speaker measurements.

If you can view and parse their info and remarks within this context, then they can be somewhat helpful and more interesting. Newbies to these subjects will probably be somewhat less able to do that though. So I'd encourage some of the newer members here to check out some of Dan's other interviews and discussions, and to compare those (and his comments in the above video) to the info and opinions of other experts and developers in the field.
 
Last edited:

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,476
Likes
4,094
Location
Pacific Northwest
I particularly enjoyed his willingness to talk about differences we can hear that are hard to measure (from 28:00), especially coming from someone like Dan who understands the value of measurement and applies it in his products.
 

MRC01

Major Contributor
Joined
Feb 5, 2019
Messages
3,476
Likes
4,094
Location
Pacific Northwest
23:14 ff. : 5% distortion in the midrange; nice implicit reference to Grado headphones :).
Yep. And his reference to one other headphone with equally low distortion that happens to be a planar, must have been Audeze.
 

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
It isn't difficult to see the relationships between the measured in-room responses of neutral loudspeakers and neutral headphones btw, if you simply compare the diffuse sound power responses of the speakers to the diffuse field compensated responses of the headphones or Harman's over-ear headphone target response curve.

This will probably work best with the diffuse field measurements made on the new HBK 5128 system, or other new rigs with a mannikin. But it can also be seen fairly easily in the diffuse field plots of neutral headphones made on some current 711-based rigs, like the ones in Oratory's or Crinacle's graphing tools, for example (which were made with flat plate rigs). And comparing them to the sound power curves in the spinorama plots of neutral speakers here, or in Pierre Aubert's database, or EAC. (Crin keeps his diffuse field headphone graphs behind a paywall though.)

This is something that Dan touched on briefly at the beginning of his presentation. But then seemed to back a little away from at the end when he was asked by an audience member about other possible target compensation curves than the Harman curve.

There are some differences and nuances in the measured responses of the headphones, particularly in the treble frequencies, that also need to be taken into account when making these kinds of SP to DF comparisons though. Especially on certain kinds of rigs. So they can't always be viewed as a precise one-to-one correlation, esp. in higher frequencies.

The lack of any in-ear measurements of the in situ/steady-state response of loudspeakers was, imo, also a significant shortcoming in his presentation, since these would probably have been very useful for comparison with some of his raw/uncompensated in-ear plots of the headphones' frequency response. This is especially lamentable given the amount of time and energy he spent in the presentation trying to draw correlations or comparisons between the two. Maybe this was info he wanted to protect for proprietary reasons though.
 
Last edited:
  • Like
Reactions: jae

ADU

Major Contributor
Joined
Jul 29, 2021
Messages
1,587
Likes
1,086
Good stuff. More focus on issues beyond frequency response is needed, as per his slide in the video. Distortion measured as a % THD (vs db below which is harder, for me, to appreciate) at various levels including peaks that correspond to 90db at a 10' or so listening distance would be easy (Erin does this at times like on his f226be review). More analysis of impulse response / decay time / ringing. etc per Dan's bullet list. In short, measure what's important not just what's easy to measure.
Indeed, the open issues segment was enlightening, including the audience comment that VR/AR and spatial audio concepts may drive further research. NIST could be leading the charge if only there was a national security or competitive aspect…
At 26:37. Soundstage. imaging. We can heard definitely we can measure. Just nobody will touch what kind of measure unit, it is good or bad.

+1 on all of the above as well. I think there are probably some audiophiles and enthusiasts who would rather not see some of these (seemingly?) more subjective characteristics reduced to simple, cold, hard measurable stats though.
 
Last edited:

robpbg

Member
Joined
Mar 26, 2021
Messages
18
Likes
21
I don't know who you are referring to but I already show headphone distortion in percentage:

index.php


And it is shown at three different levels so you can see the progression. I don't know what you mean by distance though. These are headphones so distance is constant.
The above graph is the ideal presentation in my view... I wish everyone would standardize on it. I was referring to speaker measurements needing to use a similar graph, but adjusted so that the db at one meter (typical measurement) is converted to db at 10' in a typical listening room. I am not sure but I believe there is research showing an adjustment of 10 or 12 db ish. Given that a lively presentation probably requires 90db peaks (not avg) at listening position, that implies needing to measure distortion at at least 100db/1 meter. THX spec for front LR is 112bd/1 meter apparently. So 100 or 102 seems like a reasonable requirement. Would be nice to have curves that go to at least 105db/1 meter or until compression hits.

Even better would be to offer a 2nd graph in addition, like the above, but with 12db/octave slope applied below 80hz so can easily visualize distortion with typical sub setup.
 

nate

Member
Joined
Jun 7, 2020
Messages
8
Likes
1
What do you think are the other 2 headphones of the 4 he mentioned made by Harman that follow the curve? Could they be the k371 and k361?
 
Top Bottom