Neutral all around! (Patreon)

Sorry guys but I was still busy as crazy the last days simply trying to understand how to achieve a perfectly neutral frequency response for a headphone towards my ear but realised that there may be more solutions to this problem which gave me sleepless nights as I have been continuously trying to improve and approximate this perfectly neutral final response and while I manage to improve some aspects it may worsen others.

The starting point is to have a real neutral reference. Like described before, listening to speakers doesn’t produce a homogenous spectrum especially depending on where they are listened to thus the listening room etc will influence the naturalism of the spectrum on its own just like your position between the speakers will add interferences with the “phantom center not sounding “correct” in a sense that it’s not really coming from front, because your human ear will perceive sound from front with a different response than from the sides namely from the positions where your speakers are standing, thus listening to pink noise through speakers you will end up with a fake spectrum which just doesn’t sound real because it comes from 2 single sound sources and introduces too many other problems and variables.

Therefore the only way to approximate this problem precisely was throughout my binaural recordings which in the end should sound exactly the same like I heard them in real life over a properly corrected headphone. The more neutral I manage to achieve the captured spectrum, the better the final result should be, at least that’s what I thought. The topic is way more complex and I am currently trying to understand and experiment with some factors and maybe try to deduct some rules for proper equalisings which may even improve spatial imaging just by adjusting the EQ differently for the center and the sides and I am currently looking for some ways to achieve this. I am not a scientist neither a programmer therefore I will also need help from you if you know some more sophisticated apps with more sophisticated approach to signal processing including creating filter macros with several filters included being able to adjust them at once with the correct gains relatively to each other or quickly mirroring filters etc etc. Most of these tools are completely missing from all the programs I have tried including Ozone, Audacity, Fab Filter Pro. Thus I am mostly jumping back and forth between different programs depending on what I try to achieve and mostly wasting time to prepare the audio files first because this app can only do FFT analysis on entire files, that app cannot cut files etc. It’s all very limiting and discouraging.

On top of all this my main microphones broke again just yesterday. They served me well for more than a year I think but finally the wire broke at one side which is probably not that easy to fix for an average Joe like me because the wires and microphones are so small that I wouldn’t even know how to proceed.

This prevents me of doing any further recordings right now but I am glad that I used the last days to do further recordings of various spectrums around me also with most other binaural microphones I own which will allow me to match all responses to each other being also able to use my artificial head more often than before because now I can be sure that the output should be valid as well.

I would like to invest some more time into all this and and also some money to get further pairs of the same microphones to be produced on demand by Mad Economist as I think his design is by far the best I tried although definitely not cheap.

It will also need some work to sort all my recordings and find the right snippets from all of them representing some “smooth, natural” spectrum which could then be used as base for a final headphone EQ.

I already have a first EQ for my HD600 which I will post tomorrow after getting some sleep first because I need to also clean it up a bit. I think it approaches some kind of “neutral” at least for me and sounds close to transparent to me already now although it may still need some tweaking. The result is way better than any other EQ I achieved so far. When switching from this EQ to another headphone, they all sound broken to me now. It’s just a matter of getting used to a particular sound and if these sound is as close as to that you usually would exepecg from real life, it will be hard to improve it even more except maybe compensating for some aspects of hearing loss.

You will find all my EQs in the enthusiast level. These EQs are a pain in the ass to create and require lots of effort and listening from my side to validate and optimise the results, therefore I cannot offer them for free as it costs me lots of time to get something which I personally would be satisfied with. But once I settle for a particular response I would like to do the same for most other popular headphones I own and offer you a kind of unique “correct” sound settings given that you will perceive them similarly as me. It’s not enough trying to match 2 response curves to each other based on some graph with a single click like many are offering with their EQs for free eg. Oratory1990, AutoEQ etc. These methods can be helpful to at least get a first starting point for further adjustments done through dreary AB analysis trying to hear and finally decide which one is “better” or shall I say “more correct”. That’s also the biggest problem with all this: what I may perceive as “correct” may sound completely off to someone else’s ear, due to different HRTFs. I am still not sure how well these sound settings I achieved with my ears are transferable to others and would therefore also welcome any feedback from you to know how others hear them. Some obviously perceive sound in a very similar way like me as it would seem from all the feedback I got so far; but at the same time getting to this final optimum is going to rely on the finest adjustments trying to correct for some 0.2dB bumps which in the end will define the level of “correctness”, thus once a sound appears off to someone else to begin with, those tiny adjustments needed for myself to get an even more convincing end result may not quite matter or be hearable to those. Therefore spending money on different headphones again and again with the hope one fill finally offer that ultimate experience, is a pipe dream and a simple marketing plan. With all the experience I did until today and I hardly manage to achieve this goal for myself I can surely state that nobody can tune a headphone properly so that it would sound correct to everyone, it’s even worse as those tuning targets like Harman mostly based on some random user preference may give satisfying results over a large amount of various especially untrained listeners but it will never give YOU those great results everyone is expecting. While my approach which is based on hearing with my own ears may sound not as correct to someone else as it does to me but probably it will be still better than all those more generalised and more averaged targets done with averaged silicone ear models of some super expensive artificial heads simply relying on some final curve. In the end I think that despite my efforts you will need to add some own further work trying to optimise further values on your own, not even mentioning aspects like sample variation between different units, which is one of the biggest hurdles to offer the same neutral sound to others.

And this is another point I wanted to mention namely how differently “neutral” can be approached: So far I was concentrating on sound coming from front as this is the way how we listen to music normally, with speakers standing in front. Thus the neutral or shall I say natural spectrum I was trying to use as reference for my tunings was mostly concentrated on sound coming from front. But I noticed some problem with my binaural recordings thanks to this. I can nicely equalise my recordings towards this particular spectrum using this in a next step for creating a final headphone EQ which in the optimum scenario should be able to reproduce this spectrum the same way I heard it. This way though sound coming from the sides or above me etc will create peaks in the spectrum (because the frequency response hitting your ear drum from another angle will change and emphasise particular freqencies mostly in the 6.5k and 8k range) which makes the overall spectrum get more scruffy and not as smooth which is also the reason why most headphones cannot reproduce this range really well and will rather appear harsh. Because of this there is not a single solution to this problem but neutral can be interpreted in various ways, based on how the basic spectrum was achieved and from which angles the incoming sound was mostly contributing. I am also trying to perform further tests averaging the final spectrum over more angles which I did by simply turning around while listening to the wind noise through my window being protected from wind at the same time. I looped one part so that you can hear it repeated as it’s quite a fascinating experience, although it was already past 3 o’clock AM there was still some fucking plane approaching disturbing my recording. Unfortunately I couldn’t do more because I noticed afterwards what my microphones already broke: sound all around

My plan is to also do further equalisings using an averaged spectrum over all angles, this will result in a different and brighter sound, probably more diffuse field like as the initial spectrum will be smoother, you simply iron out the frequency peaks created by the incoming sound from the other angles, therefore the headphone must be tuned brighter to be able to reproduce these peaks properly. It’s quite an interesting topic hearing these different tunings which all sound different but still plausible on their own. My current HD600 EQ is only optimised for sound coming from front, I haven’t tried the averaged one with sound coming from all around but based on all these findings I can hopefully deduct some rules how to change between a more frontally tuned sound and one samples from all around or even get some bias from above (sound from below sounds quite hollow and ugly though.)

Having this option available through some app, and being able to switch between different tunings ratios, settings etc would be my dream. Offering you a product with all these options and maybe even other personal tuning adjustments to taste with no claims of being correct, would be something I would definitely love to do. Although right now I am still trying to find a valid “neutral” kind of sound first, because without having this and knowing how this sounds, one cannot judge sound properly. Your brain will immediately “break in” after a while to the given sound, thus it’s not that easy to find the “correct” ones as there will be various was to interpret neutral just like I described.

I also wanted to demonstrate you the problem with these different approaches to tuning with snippets from 2 of my previous binaural recordings which I posted before. I think meanwhile I managed a better final compensation for my binaural recordings compared to before and would like to show you the difference in sound also asking you which you perceive as more natural or more correct sounding, maybe some can already listen with my new HD600 EQ because this should give the most correct representation. I will post the wind and rain example from last time again and you will probably notice that they both sound different (and hopefully more correct than those older samples). First is with the sound equalised to a spectrum coming mostly from front, while the other one is equalised to more averaged spectrum from all around. After a while of listening your brain will get used to any of these, but I can already tell you that with my current EQ for the HD600 the frontal one sounds definitely more correct despite appearing harsher compared to the other one also imaging should appear quite different on both. But the rain example is extremely hard to render faithfully it may be also the limit of what kind of realism is achievable with binaural recordings at all.

wind correct

wind averaged

rain correct

rain averaged

Having some final correct “responses” which I can measure or somehow capture will allow me to transfer these to other products and hopefully also IEMs which will be harder to achieve though as I cannot measure them properly with my couplers and need to approach this optimum curve mostly by listening.

I just want to thank everyone who is supporting me with all this again because I finally I found some sense of life again trying to figure out why this sounds like it sounds and how it can be improved. The most interesting experiment would finally be to achieve the same response with 2 different headphones as close as possible and check if other sound aspects may come into play justifying a 10 times higher price for one or the other!