• 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.

Headphone Correction & Spatialization

andreasmaaan

Major Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
4,845
Likes
4,936
#1
Hi all,

I recently went about doing all I could to improve the sound of my pair of HD6XX. I thought I'd share here what I've learnt.

Objectives

Having found myself unexpectedly stranded in Australia due to covid19, and without headphones or access to a decent speaker system, I bought myself a pair of HD6XX a couple of weeks ago. I've never been a big fan of headphones, but now that I'm here for at least another 6 weeks, I'd started to miss high-quality audio. I also dabble in mixing/mastering music, and have been missing that too while down here. I've previously used headphones for low-stakes mixing (e.g. of audio for videos intended for online publication).

Anyway, I have a bit of time on my hands, so I decided to see what I could do to get these headphones sounding as good as possible, with a secondary goal of providing a neutral reference for basic mixing.

Part 1 - Amplitude Response

Fortunately, there are plenty of measurements of these headphones (or the acoustically identical HD650) available online. I settled on these measurements as my starting point, as they were taken on GRAS43AC/43AG:

1589204427245.png


As a first step, I used Equalizer APO to plug in the suggested EQ settings in the link above, in theory bringing the response to meet the 2018 Harman target curve.

1589204537500.png


That did subjectively improve the sound somewhat, especially in the treble and the upper bass (as can be seen above, the HD6XX's treble is a bit recessed and there's bump around 200Hz with a gradual roll-off below).

Still, I didn't feel that this target quite hit the mark for me. The bass seemed overwhelming (I've always found this with the Harman target curves), and the treble still sounded too mellow compared to my perception of neutral.

Next, I repeated the same steps, but this time aimed for a flat diffuse field target. This graph (taken from Olive) shows the Harman 2013 headphone target (black) overlayed with (what is for all intents and purposes) a flat diffuse field target (green):

1589204817768.png


That sounded a lot closer to the mark to my ears. The treble was less forgiving on some poorer recordings, but to my ears this was more neutral/correct, and on good recordings, more satisfying. The bass was definitely a lot more restrained though (too restrained for many recordings). Overall, for almost all recordings I preferred a diffuse field treble response, but for some recordings the bass seemed lacking.

I ultimately settled on a frequency response that was roughly in between the two. I had switched to Ableton Live by this point instead of Equalizer APO. This is a screenshot of the EQ settings that I found most suitable for the widest range of music. It's close to a diffuse field response above 200Hz, with an effective bass boost of around 3 or 4dB below 200Hz.

1589205436267.png


Part 2 - Spatial Enhancement

But still I wasn't happy. I've always found the in-your-head sensation with headphones to be pretty unpleasant. I don't often use headphones for this reason. I decided to try out some spatialization plugins.

Of the plugins I tried, Waves NX and Ghz Can Opener were the only two I could find that had a net positive subjective effect for me.

1. Can Opener

With Can Opener, the effect using the default settings was not dramatic, but the subtle improvement was a good one. At the default 60°, the stereo image moved out of my head quite a lot at the L/R extremes, and far enough out in front in the centre to relieve me of the in-head feeling I mentioned earlier. With the angle set to 75°, the effect was even more impressive at the extremes, although I thought just slightly less convincing in the centre.

In the end I settled for the settings shown here:

1589207211317.png


The only discernible negative effect I found was a slight perceived thinning of the tonal balance. I investigated this by using a spectrum analyser to compare the spectrum of the output of averaged white noise with and without the effect engaged.

Sure enough, with the plug-in engaged, the spectrum went from this:

1589207614313.png


to this:

1589207558100.png


That is also what it subjectively sounded like: thin and recessed in the midrange.

I then applied additional EQ to correct the output from the plugin so that it matched the pre-plugin amplitude (note that the headphone correction EQ I'd previously settled on remained engaged here at all times, but remained downstream from the spectrum analyser in the signal chain).

1589207772402.png


Subjectively, with both white noise and music, the spectral balance now sounded identical whether the plugin was engaged (with EQ) or not, with the spatial qualities of the plugin still fully effective.

This is actually more than I'd hoped for :) Preserved tonal balance combined with a subtle but pleasant shifting of the image outside of the head, and with the added bonus of a really nice sense of width and depth at the L and R extremes. On the other hand, a little disappointing in terms of pushing the phantom centre image out in front.

2. Waves NX

Next, I had a play with Waves NX. This is clearly a more sophisticated plugin, with configurable settings for head geometry, plus a webcam-based head tracker (a hardware tracker can also be purchased).

1589208969770.png



I plugged in my head measurements and (initially) disabled the tracker to see how it sounded.

Subjectively, my first impression was that it was a lot less subtle than Can Opener, with the image shifting significantly further out in front of the head, but with less overall spaciousness (particularly for hard L/R panned sounds), and a greater sense that the spectral balance had been disturbed. Overall, I would describe the sound of NX as more akin to a pair of speakers in a room (even with the "room ambience" setting on zero).

However, it also seemed less tonally accurate to the source, with the highs instantly seeming sharper/harsher.

Again running white noise through the plugin and analysing the spectrum, this is what I found:

1589208347658.png


With head-tracking engaged, it was clear that this FR was highly dependent on head orientation, and very different in the L vs R channels except when the head was directly in the "sweet spot". I didn't take screen shots at all the different angles, but it seems clear that these curves are based on real measured response curves taken at various azimuth/angles.

Impressively, the response changed with both horizontal and vertical changes in head orientation.

It was difficult to know how this could/should be equalized. I couldn't tolerate the perceived harshness with the plugin engaged, but it was also quite obvious that the spatial effect was more impressive than that of Can Opener.

After playing round for some time, I found this curve to mitigate the perceived sharpness I'd been hearing at first and to bring the tonal balance closer to what it was without the plugin. Some more fiddling around will be needed though.

1589213696574.png


I then played a bit with the head tracking, which actually works very well, giving pretty plausible subjective results to +/- 90° horizontally and maybe +/45° vertically. This significantly helped to externalise the stereo image. Unfortunately though, the lowest latency I can get from this computer with system audio routed through the plugin chain is just under 30ms. This is enough delay to make the effect a little disconcerting.

EDIT: I forgot earlier to post the impulse responses.

No processing (soundcard loopback):

1589244639579.png


Can Opener (with settings pictured above):

1589244694188.png


And Waves NX (also with settings pictured above):

1589247510106.png


Summing Up

Overall, I found further confirmation that the Harman curve doesn't match my idea of a smooth, neutral frequency response, coming across as overly warm/dark and too aggressive in the bass (I've previously found this with other headphones, too). However, the general curve of the Harman target, with slightly boosted bass and somewhat recessed highs relative to a flat diffuse field response, does seem to agree with me: in the end, I went for a curve somewhere in between the two.

In terms of spatialization plugins, this was a revelation to me :) I now can't imagine going back to using these headphones without one. IMHO, Can Opener is worthy for having virtually no effect (post-EQ) on perceived tonal balance while nevertheless successfully pushing the stereo image slightly outside the head. It's particularly effective for images close to the L and R extremes of the soundstage: these sit right out of the the head, and older-style over-hyped stereo recordings (think Bill Evans' Night at the Village Vanguard or most the Beatles's stereo catalogue) actually sounded really satisfying, despite the overzealous panning.

Waves NX on the other hand really pushed the frontal image out further, emulating better my sense of what speakers in a room sound like. The head tracker is very effective (although latency was an issue in my case), and really assisted the construction of an external stereo image.

On the other hand, I couldn't imagine engaging this while mixing audio - the perceived tonal balance is just too different to be relied upon. It might be useful to flick on and off occasionally to check how something might image on speakers (I couldn't AB it against speakers here to form a strong impression of whether that might be the case).

Overall, I'd strongly recommend trying both FR correction and spatialization plugins for headphones. After listening like this for some time, when I now turn off the full signal chain, the sound seems barely a shadow of itself...
 
Last edited:

Bob-23

Active Member
Joined
Jul 3, 2019
Messages
248
Likes
159
Location
Berlin, Germany
#6
Bill Evans' Night at the Village Vanguard
...one of my all time favourites! Great work of yours - I couldn't listen to headphones without crossfeed, tried out several versions, some were terrible, I, now, prefer the Jan Meier crossfeed/bass enhanced (hardware version) - and, btw, to me, the in-head vs. before-head perception also seems to have a 'psychoacoustic' component...
 

Thomas savage

Power hungry desperado
Moderator
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
9,733
Likes
13,334
Location
uk, taunton
#7
...one of my all time favourites! Great work of yours - I couldn't listen to headphones without crossfeed, tried out several versions, some were terrible, I, now, prefer the Jan Meier crossfeed/bass enhanced (hardware version) - and, btw, to me, the in-head vs. before-head perception also seems to have a 'psychoacoustic' component...
Mine too , absolutely love Bill Evans .

Brilliant post @andreasmaaan , a huge thank you for putting it together and sharing it with us .

I hope you get home soon but if being stuck gives you more time to post stuff like this maybe I don't :D
 

Theriverlethe

Senior Member
Joined
Oct 20, 2019
Messages
346
Likes
223
#8
...one of my all time favourites! Great work of yours - I couldn't listen to headphones without crossfeed, tried out several versions, some were terrible, I, now, prefer the Jan Meier crossfeed/bass enhanced (hardware version) - and, btw, to me, the in-head vs. before-head perception also seems to have a 'psychoacoustic' component...
I've wondered about this myself. The HD800's are the only headphones I've heard that consistently produced anything resembling a soundstage, even with crossfeed, DSP, etc. How much of that is due to the cognitive dissonance that I already know the sound is coming from objects on my head?
 

maverickronin

Major Contributor
Joined
Jul 19, 2018
Messages
1,316
Likes
1,259
Location
Midwest, USA
#9
've wondered about this myself. The HD800's are the only headphones I've heard that consistently produced anything resembling a soundstage, even with crossfeed, DSP, etc. How much of that is due to the cognitive dissonance that I already know the sound is coming from objects on my head?
That's definitely an element. It's basically the same as an optical illusion which snaps back and forth between two different images.



Another element is your own personal HRTF. Something like the HD800 with largish drivers, farther away from your ear, and at an angle adds some of you personal pinna transfer function to the sound giving it a better sense of distance and space.

Depending on your personal anatomy it's possible that generic crossfeed and DSP implementations won't work very well for you and you would need something with personal measurements such as Impulcifer mentioned above.
 

Bob-23

Active Member
Joined
Jul 3, 2019
Messages
248
Likes
159
Location
Berlin, Germany
#10
The HD800's are the only headphones
I've only heard them once, maybe some day I'll get some second hand ones... The K702 might be somewhere inbetween HD600 (I listen to both of them mainly) and the HD800, regarding soundstage. I like the K702 very much, think it's one of the most underestimated phones, although it belonged to the 'holy trinity', once, in ancient times. But it must be heavily equalized, and it absolutely needs high quality recordings (which you luickily can still find quite a lot in Jazz). - With regard to the soundstange: I've got the impression, that it strongely depends on the quality of the recording, too, sometimes everything's cramped into the middle of the soundstage , one instrument over the other...
 

Theriverlethe

Senior Member
Joined
Oct 20, 2019
Messages
346
Likes
223
#11
I've only heard them once, maybe some day I'll get some second hand ones... The K702 might be somewhere inbetween HD600 (I listen to both of them mainly) and the HD800, regarding soundstage. I like the K702 very much, think it's one of the most underestimated phones, although it belonged to the 'holy trinity', once, in ancient times. But it must be heavily equalized, and it absolutely needs high quality recordings (which you luickily can still find quite a lot in Jazz). - With regard to the soundstange: I've got the impression, that it strongely depends on the quality of the recording, too, sometimes everything's cramped into the middle of the soundstage , one instrument over the other...
I'm talking more about the obvious difference vs. a real stereo 7+ ft. away, or even desktop monitors. HD800 actually made me feel like there was a stable center image about a foot in front my face, if I remember correctly.
 

Bob-23

Active Member
Joined
Jul 3, 2019
Messages
248
Likes
159
Location
Berlin, Germany
#12
I'm talking more about the obvious difference vs. a real stereo 7+ ft. away, or even desktop monitors. HD800 actually made me feel like there was a stable center image about a foot in front my face, if I remember correctly.
Got it. That's what I was referring to with 'psychoacoustic component': the soundstage in front of your head vs. in your head; most of the time I don't think of it at all, and it only appears as such or as 'a problem', if you will, when concentrating on it, and then, as in maverickronin's picture, I may, sometimes, voluntarily, even switch back and forth...

The other aspect of soundstage is its width (and its depth); there the quality of the recording comes into play. Instruments cramped into the middle, one over the other, or evenly spread from the very left to the very right of the soundstage. My impression is, that more and more recordings abandon the concept of a realistic soundstage, with well placed musicians, with drums or keyboards that don't reach from the very left to the very right - but, well, it might be just a reflection of the fact that in more and more recordings the musicians never played in such a stage-/live configuration, and might even have not met at all, personally. So that the realistic 'correct' image would now rather be the illusionary one... - Nevertheless, there might be some interdependence between, at least, the depth of the soundstage and the perception of 'in-head' vs 'in-front-of head', maybe also between the width of the soundstage and the perception of distance, but I'm not sure...
 
OP
andreasmaaan

andreasmaaan

Major Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
4,845
Likes
4,936
Thread Starter #13
Got it. That's what I was referring to with 'psychoacoustic component
FWIW, what you're actually talking about here is a psychological (as opposed to psychoacoustic) component.
 
OP
andreasmaaan

andreasmaaan

Major Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
4,845
Likes
4,936
Thread Starter #14
...one of my all time favourites! Great work of yours - I couldn't listen to headphones without crossfeed, tried out several versions, some were terrible, I, now, prefer the Jan Meier crossfeed/bass enhanced (hardware version) - and, btw, to me, the in-head vs. before-head perception also seems to have a 'psychoacoustic' component...
Mine too , absolutely love Bill Evans .

Brilliant post @andreasmaaan , a huge thank you for putting it together and sharing it with us .

I hope you get home soon but if being stuck gives you more time to post stuff like this maybe I don't :D
Haha thanks both of you :)

One of my all-time favourite artists too.
 
Last edited:
OP
andreasmaaan

andreasmaaan

Major Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
4,845
Likes
4,936
Thread Starter #16
Edited the OP to include impulse responses of the spatialization plugins^^^
 
Top Bottom