• 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 Knowles Curve” For Headphones

Matias

Master Contributor
Forum Donor
Joined
Jan 1, 2019
Messages
5,029
Likes
10,797
Location
São Paulo, Brazil
OP
Matias

Matias

Master Contributor
Forum Donor
Joined
Jan 1, 2019
Messages
5,029
Likes
10,797
Location
São Paulo, Brazil

Attachments

  • preferred-response-white-paper-final.pdf
    2.7 MB · Views: 807

threni

Major Contributor
Joined
Oct 18, 2019
Messages
1,277
Likes
1,519
Location
/dev/null

edahl

Senior Member
Joined
Jan 18, 2021
Messages
398
Likes
328
That curve is sure to sell some estat drivers, haha
 

Curvature

Major Contributor
Joined
May 20, 2022
Messages
1,074
Likes
1,353
Saw something off and had to comment.

The Knowles paper didn't say how old it was. Maybe it came before the Harman IEM research since it's referencing the wrong curve. I doubt it since one of the songs used for the listening test was Ariana Grande's Breathin', released 2018.

Compared to Harman's 2017 IEM curve (looked it up online) the Knowles seems to confirm it. How different are the ear simulators and what effect would they have over 10kHz?

Harman IEM 2017 vs. Knowles (zoomed).png
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,389
Location
Seattle Area
The IEM target I use currently is the older one that is actually pretty close to their proposed one per member @Curvature above. When I equalize IEMs to that, at first the response is very pleasing. You get more detail, air, etc. But as I listen more, I worry about it being fatiguing and too much. So a couple of days ago I started to adopt the newer Harman curve without that peaking. I suggest to Knowles folks to survey people listening for longer than in a test and see if that preference lasts.

I also have as my frame of reference Dan Clark headphone which clearly has less highs than their proposed one and the one I have been using.


 

half_dog

Senior Member
Joined
Jun 28, 2018
Messages
452
Likes
361
Location
Brazil
One thing that I miss in most of the targets and even in measurements is the sub sonic behavior of the headphone. People say we "only" hear above 20Hz but there are some papers telling otherwise. And with phones (HD280PRO and Moondrop Kato in my experience) capable of playing below 20Hz I'm able to hear it (until something around 17Hz without dial up the volume) and this makes difference in songs where the producer left some content "there".
 

csglinux

Member
Joined
Apr 28, 2019
Messages
5
Likes
6
A few thoughts on this.

Firstly, it's worth bearing in mind that 'white paper' = non-peer-reviewed marketing material.

It's curious how this was deemed worthy of reporting on Forbes, Bloomberg, etc. As the OP hints at, there's a new target curve born every minute. At first glance this one looks like an independent corroboration of Harman up to 10 kHz, but it's not. Knowles just aimed to get as close as possible to Harman up to 10 kHz. They seem to have offered their listeners a limited selection of filters (filter?) in the treble. The amount of treble boost preferred being a function of age ought to have been obvious - as would the frequency location of the peak, if they'd given listeners that option, but it seems they didn't.

It looks like Knowles never got the memo about the GRAS RA0401 coupler. That coupler was designed to strongly suppress the half-wave resonance, but it's not physical to do so to that extent. Knowles' complaint that 'undamped couplers' cause a 'variation of response curve with insertion depth' is just parroting GRAS' original sales pitch for the RA0401/2. Response curves physically will (and should) vary with insertion depth. We've measured a ton of ears with RA0045 (711) and RA0401 and the latter doesn't correlate well with actual humans. In fairness, GRAS never claimed it would; RA0401 was just designed to be more 'repeatable'. (Flat lines are also very repeatable, but don't tell you much about headphone response.) The Harman curve was produced with standard 711 couplers. For most IEMs (and especially earbuds), RA0401 will not agree with 711 measurements, even up to 10 kHz. Most curves out there are still produced with standard 711 couplers, and those response curves shouldn't be compared against (or EQ'd to) this target curve.

It looks like a kindergartner wired up their mems mic in Fig. 3 :) The mic is supposed to go INSIDE the eartip to properly check for seal.

Most of us are not listening to binaural mic recordings from dummy heads. We're listening to recordings from stage/studio mics where the high frequencies have had no distance to travel and no time to dissipate - they're effectively getting a shortcut directly into our ears. And then Knowles are boosting them. My purely subjective take on their curve - I tend to find the original Harman curve to be already too bright. Amirm's point is a good one: a short test listening intently for detail is not the same as kicking back and listening to music for an extended period. The Harman metric involves two parts - a delta error to the target curve, and also an error based on overall difference in slope. Bumping up the treble like this is tilting that slope. I have a feeling many are going to find this too bright for regular listening. But maybe 'many' is still less than 50%, in which case it may still be $$$$ for Knowles sales dept. ¯\_(ツ)_/¯

There could be something in this, but it feels more like Knowles are just preparing some hype for their upcoming 'hi-res' true-wireless earbuds.
 
Last edited:
OP
Matias

Matias

Master Contributor
Forum Donor
Joined
Jan 1, 2019
Messages
5,029
Likes
10,797
Location
São Paulo, Brazil
standards.png
 

MayaTlab

Addicted to Fun and Learning
Joined
Aug 15, 2020
Messages
924
Likes
1,512
It looks like a kindergartner wired up their mems mic in Fig. 3 :) The mic is supposed to go INSIDE the eartip to properly check for seal.

Could it be the case that the MEMS' opening is linked to the front volume via an internal duct and the PCB sealed with glue against the body ?
Screenshot 2022-05-23 at 21.54.39.png
 

sejarzo

Addicted to Fun and Learning
Forum Donor
Joined
Feb 27, 2018
Messages
961
Likes
1,066
Most of us are not listening to binaural mic recordings from dummy heads. We're listening to recordings from stage/studio mics where the high frequencies have had no distance to travel and no time to dissipate - they're effectively getting a shortcut directly into our ears. And then Knowles are boosting them. My purely subjective take on their curve - I tend to find the original Harman curve to be already too bright. Amirm's point is a good one: a short test listening intently for detail is not the same as kicking back and listening to music for an extended period. The Harman metric involves two parts - a delta error to the target curve, and also an error based on overall difference in slope. Bumping up the treble like this is tilting that slope.

Thanks for this because I could not agree more strongly. Seems analogous to the oft-stated impression that listening to studio monitors in the nearfield with a near-flat response at the LP makes sense for production but not for consumption, as it were.

I'm 65. For the past 10+ years I have known that I have a 6-8kHZ dip in my hearing plus some tinnitus. Using a variety of IEMs and the sweep function in MathAudio Headphone EQ, I still can hear up to at least 13 kHz--barely though, as I sense a drop off just below 10 kHz. While that might seem to mean I should prefer at least the same treble boost that either target suggests, nothing could be further from the truth.

If I EQ any headphone or IEM to Harman target via convolution applied in my Foobar DSP stack, I almost always add in MathAudio Headphone EQ with a high shelf cut of -1.5 to -2.5 dB/~0.6/1500 Hz on any close mic'ed material. At times, I need an additional -2 dB peak cut from 2-4 kHz.

Classical orchestral or wind band recordings that I know were minimally mic'ed with just a stereo pair are about the only ones that sound properly balanced to me with headphones or IEMs EQ'ed to Harman target.
 
Top Bottom