oluvsgadgets
patreon
Oluv's "neutral" & "universal" headphone targets
🕑 Added 2023-03-30 11:05:30 +0000 UTCWhen I am deep into something, I can hardly concentrate on other things, therefore please apologize my recent silence but I have been struggling lately to finally settle my own headphone target and come up with a sound profile that should render my own sound samples correctly and appear to some degree enjoyable for usual music listening. It would seem as if many audio enthusiasts weren't that satisfied with the common targets from Harman research and are trying to develop their own like seen here.
I have tried Resolve’s EQs among all other common EQs by Oratory1990 and other accessible AutoEQ presets I could find which all sounded off to me. Therefore I put lots of energy into hopefully a target which could not only satisfy me but also others. As audio is not an exact science but you have to deal with lots of psychoacoustic effects besides the brain tends to adjust to the given sound pretty quickly that’s why it took me again way longer than I hoped. What I thought was great the first day, sounded weird to me the next day, I was drifting away too much from my original rough attempt that you can find here with my methodology described more in detail here creating dozens of variants but in the end I figured out they all didn’t sound correct, although after some listening the brain immediately accepted them, but I aimed for a truly valid and correct kind of target.
After many trials & errors, I figured out that there was simply no way to manage a unique all-embracing headphone target which would be the holy grail of targets. If I managed to make my recordings sound as real as possible, the headphone sounded overly bright and harsh with little bass especially if the recordings were compensated to be strictly neutral. I had to find a compromise and finally I decied to go with 2 different targets, because it was impossible for me to create "neutral" binaural recordings, equalize a headphone to render those correctly and still hope this headphone would sound decent per se. Therefore I went back to my old target I derived from my wind recordings you can find here and found it to sound actually more "neutral" when compared to the sound of speakers in front of me with just the overall balance being off a bit and tried to readjust it according to my latest findings and measurements and I think I managed a nicely "neutral" sounding target, which may still need some slight adjustment in future, when I get a better neutral reference. A friend of mine wants to sell his Neumann KH80 studio monitors. My first attept to purchase a used pair from some online seller one year back ended in becoming a fraud victim, but this time I hope I can finally get these during the next weeks and use them to have an even more valid neutral reference, because without a continuous comparison to a real reference any adjustment may depart more and more from the actual target. During the last days I managed to settle this target, or let's call it EQ for my HD600 and smoothen it to such a degree, that I don't perceive anything annyoing anymore. This will be the "neutral" target, which basically should make music sound as if you were listening to a neutral set of speakers (without the room influence added). I didn't bother with bass-shelf etc to simulate the in-room response like Harman was trying, because IMO this makes recordings appear way too bassheavy, I rather wanted to hear the "original recording" with as little coloration as possible. I already managed to bring down EQ amount down to 15 filters and also optimise the filters to work for the Qudelix 5K because the Qudelix always sounds different than EqualizerAPO with exactly the same values for no obvious reason. With the recent firmware update the Qudelix 5K supports up to 20 filters, or 10 filters per channel, which can allow for separate equalising for both ears, but I prefer having more filters than less per side.
The other target I would maybe like to call "universal" or "precise"? I am not sure, maybe you can give me some hints. It is derived from my latest spectrum and compensation adjustment to render my recordings as faithfully as possible. Listening to my recordings with a headphone equalised to this target should give you the most realistic results I was able to achieve so far. I tried hard to get it even closer, and I am not sure I can manage even better, there are limits how faithful the "real life" can be reproduced over headphones and I am not even sure how close this result will be for others. If I can hardly manage it for myself properly, I wonder how this could even work for other ears, let alone headphones where every unit of the same model sounds different due to sample variation etc...
The "universal target" still sounds "correct" though, not only with my own recordings but also with music etc, it's just a different interpretation of the HRTFs. If you listen to pink noise from front or from the side it will still sound correct for you although the actual response will be completely different. Here the difference of the spectrum I captured with mostly frontal incidence compared to that with additional side incidence:
The difference of purely side and front spectrum would be even bigger, but those effects are included within the "universal target" which sounds more spacious as it has more of the side spectrum included, therefore it's also brighter than "neutral", so compared to a speaker it will appear different, although the character of the spectrum is still retained to some degree. It resembles more the sound of the HD800, but of course much smoother and "more correct". Which target you prefer will depend a lot on your taste, but after some minutes of adjustment I think that the "universal target" is definitely more enjoyable and gives you more of an immersive feeling.
To put this into some perspective I would like to show some graphs. Keep in mind you cannot compare them to standardised HATS measurements, because I am measuring with my own ear at the ear entrance point and not at the ear drum, but I can show you at least how the HD600 measures stock (grey) and with my new neutral target (black):
Here against my universal target:
Here you can see my old "neutral" in grey compared to the new targets with a stronger 1/2 of an octave smoothing to demonstrate the differences in the overall balance better. Basically just the balance above 1khz differs between them all. But simply raising that area doesn't work as you need to readjust all critical frequencies to get the same decent smoothness.
Next you can see the universal target compared to the frontal spectrum I measured from the speakers equally with my ears:
Notice how the target aligns to the reference spectrum above 2khz with way more dip at 1.5k and 3.5k than any other target out there, but this seems to be more inline to what my ears perceive as natural. I also had variants where I tried to stay closer to the original spectrum by reducing additional peaks, this resulated in a smoother perceived sound, but also rendered my own recordings less realistic and more dull:
Finding the right balance between smoothness and realism was actually the hardest part which took me the longest to achieve decent results. I personally perceive the neutral target as smoother and more natural and also more similar to what I hear from the speakers, although the universal taret aligns better with the original spectrum and also renders my recordings more faithfully which are also strictly equalized to the original reference spectrum. It is still beyond my understanding why this deviation is happening, also why the headphone target measures with more bass than the real speaker spectrum, while actually it is perceived less bassheavy. I am sure there are scientific papers on those topics, still I wonder why nobody has managed a real faithfully sounding headphone yet. Although the HD600 is definitely one of the more neutral ones I have heard, you can see that I depart a lot from its response making it sound quite off without any adjustment, at least to my own ears. Let alone headphones which are
I would definitely like to optimise everything once I have a better neutral reference, maybe I can somehow merge both my target approaches and finally manage a unique all-embracing headphone target once hahaha....
What does this mean for you? In order to listen to my recordings with the correct response, I will share my recent raw and unedited HD600 EQ for my "universal target" here so that all my Patrons can at least have an option to listen to my sound samples how they are meant to be listened. The EQ is still a bit scruffy sounding, but it should have improved a lot over my last attempt I posted a while back, especially my recordings should sound way more real now. All HD600 owners, please let me know how you perceive the EQ with my own recordings, but also with music. You can find many binaural real life samples on my 3d binaural channel. My shorts contain lots of recordings with complex natural spectrums which you can use for judgment how "real" they actually sound. I welcome any feedback.
My plan is to optimise this EQ in future, maybe I will manage to make it sound smoother while still maintaining or even improving the rendering of my recordings and also reducing filter amount to make it usable for the Qudelix 5K. Then I will offer this EQ together with the neutral EQ for the HD600 to all my Patrons within the enthusiast-level only as well as all upcoming headphone EQs in both variants, neutral and universal in both EqualizerAPO and Qudelix 5K versions.
It is hell of a work to get decent results, therefore everyone taking the enthusiast-level will get a new headphone EQ every month, this will be my main "occupation" to make exisiting heapdhones sound better also improving on my target continuously. Upcoming headphones will be HD800S, HD560S, DT770, DT880, DT990 etc. I also plan to select a cheap headphone next that everyone can afford and offer my universal EQ for this so that everyone can listen to my sound samples correctly. I already started but notcied that it needs way more effort than just aliging to graphs. 2 identical measurements from 2 different headphones can still sound completely different.
My biggest dream would still be to get the chance to work on a wireless headphone once including all my sound profiles together with additional HRTF front/side spectrum compensations, which you could control by a simple slider and maybe even including simulations of exisiting popular headphones. But I hardly doubt this will ever happen. I lack the real contacts, and nobody really seems to be interested. Everyone tries to cook his own soup with the given results we all know.
I would like to thank again everyone for your continuous support. You allowed me to waste my time with all this with pretty little benefit actually, at least according to my wife. Sleepless hours of listening, tweaking, listening again the next day just to discover that the work of 2 days became obsolete, when noticing that my earlier adjustments were actually better. In real life nobody would pay for such a work. It's the final result that counts and how much sales it can generate. How it sounds, doesn't really seem to matter. I am still baffled that despite my sound samples many still argue that the doubtlessly worse sounding headphone sounds better to them.
This is the only contribution I can offer to the audio community which can hopefully be of some value to others too. As long as I am still getting support from you I can hopefully waste my time with this a bit longer although nobody can tell for how long. A public video explaining what I was trying to achieve will be coming soon.
PS. sorry, I had uploaded the wrong EqualizerAPO file before, as I am already confused with all those different versions. You can download the correct one now. I also updated the graphs with the correct versions.
Comments
Adding the dip at 3.5k seems pleasant to my ear as well
Oluv'Sone
I can only judge what I am hearing in real life when hearing pink noise over speakers or with my current target, both are way closer than they were when I didn’t do those adjustments which I have now. Not only 1.5k but also 3.5k seem to need some dip in order to be perceived natural.
I’ve noticed good results using IEMs and having less energy in the 1.5khz area. Appreciate your find
Good work.... Enjoying your efforts
