How does „neutral“ sound like? (Patreon)

I bet nobody has really heard a “neutral” sound out of a headphone yet. Despite all claims: this headphone is the most neutral one, or that one, from my own experience no headphone or IEM I have heard so far was really close to offer a fully neutral sound reproduction. I must confess that I had also relied too much on existing established standards and assumptions, thinking Sonarworks or other targets and equalisings may offer any “true” sound, without having them fully verified by myself.

This is exactly what I tried to achieve now completely from scratch and from behind or the other way round, because this time I started with my own binaural recordings as base to equalise my headphone to and finally get the same sound I am usually hearing in real life. So far I had equalised my headphones towards pink noise played through some neutral monitors using an OptimumHifi profile as base for my own corrections and adjustments. This introduced lots of errors as pink noise played over speakers is never really flat except when listened in an anechoic chamber maybe. Thus my achieved headphone EQ was not correct to begin with while my binaural recordings equalised towards this profile which I had achieved assuming it to be “neutral” carried on with the same errors introducing new ones. I had 2 errors to deal with: one is the equalisation error of the headphone towards some speakers, the other one is the error by equalising my recordings towards the headphone which already was flawed. Because of this neither my recordings were sounding correct nor my allegedly neutral headphone was really neutral.

But my goal was to get it as correct and solid as possible at least for myself to finally be sure that the sound I am listening to is really “correct” and neutral. Only then I can have a true neutral reference which would allow me to judge other headphones relatively to this in a correct manner without any bias.

I wanted to describe my current workflow which I think managed to give me the best results so far regarding both the quality of my binaural recordings and the neutrality of my headphone.

First I had to equalise my recordings towards some neutral reference like described already before. Pure pink noise does not exist in nature thus the only way to get some pink noise for equalising would be to play it over some speakers which will introduce their own flaws and errors because of various factors influenced by the listening room, the listening position, positioning of the speakers etc.

To overcome this problem I used natural wind noises, rain etc as my reference spectrum which is even with a smooth frequency response sounding way more pleasant and natural than pink noise. To get an increased front bias and avoid too much sound hitting my ears from the sides or back I did several recordings from inside my room with an open window limiting the horizontal angle of incoming sound to about 120° or simply standing outdoors leaning at wall behind me or laying at the ground and recording upwards, which would concentrate sound emission to only come from front.

Next I tried to match the responses of my own binaural recordings to the same spectrum recorded with normal microphones to avoid including any frequency anomalies thanks to my own HRTFs captured. I used different recordings and tried to create an approximate average response which would align best over all other reference spectra I recorded with no frequencies standing out in any of them.

Finally I tried to achieve an EQ for my headphone which would render my equalised binaural recordings as faithfully as possible to the same degree like I perceive these noises in reality. Thus hearing the recording of wind over my headphone should sound nearly the same as the wind I am hearing in real life. I attached 2 binaural example recordings for you to check how you will perceive them. One is an arriving storm with lots of wind, the other one some heavy rain, both recorded in front of my door. You should mostly hear the sound to the front around you as behind me was a wall in both cases. I was standing with some distance to the wall to avoid too much interference with the reflected sound coming from behind. You can find the uncompressed WAV files here:

Wind: https://drive.google.com/file/d/1je-hBwxqH6OV7KqST7WFEGyLRSbCxOR6/view?usp=sharing

Rain: https://drive.google.com/file/d/1hR3wI7k4wu0StWZcKVCCAIMDuxV1qRZ8/view?usp=sharing

Please check how realistic these recordings will sound over your headphone but they will probably sound off regardless which headphone you use. If you listen to them over neutral speakers the sound should be pretty realistic with no obvious anomalies in the spectrum though. The problem here is simply the common headphone sound, which obviously does not manage to reproduce a natural spectrum correctly.

To my surprise the final EQ for my HD600 deviates quite a bit compared to my initial one, sounding closer to a HD800S now. Many claim that the HD800S is strongly tuned towards diffuse field which would indicate that my EQ is also more diffuse field like although limited to the range above 2khz otherwise it would sound much brighter. Listening to my corrected recordings with the new setting I can still hear some slight ringing in the upper frequencies which I would like to correct for if I still manage, but as soon as I adjust something it may worsen other parts, thus the entire procedure is quite tricky. Nevertheless the sound I achieved right now is already way more natural and smoother sounding than all my previous attempts not only when listening to my binaural recordings but also to pink noise or normal music, it sounds way better and transparent now. The HD600 without any EQ sounds really broken to me now.

Here you can see my old EQ curve for the HD600 in grey which was also used as base for my reference tuning of the Free Pro 2 overlaid with the new EQ curve in black. The area above 2khz is mostly reduced compared to my previous setting, which makes mids and treble sound more natural than before.

The next step will be to recreate this response for the Free Pro 2 as good as possible to allow for an equally faithful reproduction of my own binaural recordings but also music in general. I would like to make an entirely new tuning out of this prepared for various ear canal resonances just like I did before. The only problem will be sample variation which can be pretty high between all units I tried so far. Thus despite my strongest attempts to offer you an as correct as possible response towards your ear canal resonance there are still points of uncertainty because of sample variation and your own outer ear which may simply differ way more from mine and expect a different kind of response to be perceived as natural than me. I am not sure how well my current setting may be “generalised” although with the experience of most of you here, many seem to have got decent results with my previous settings so far.

I will also need help from others as I am looking for a way to approximate existing EQ functions by different amounts of filters. My current EQs consist of dozens of EQs to fine tune for some details but I would like to create a final EQ needing to reduce filter amount by a high margin. I remember some of you have mentioned such tools, so please let me know if you have an idea how I could proceed. If I get a tool which would allow me to approximate filter curves with a reduced amount of filters I could create EQs for different kinds of headphones more quickly and easily than having to do it manually all the time. It could definitely speed up my workflow.

But currently I am still trying to optimise my current EQ for the HD600 and validate it with some further listening tests until I can start working on a new tuning for the Free Pro 2. Getting a final valid response curve for IEMs which is as close to neural as possible will also help me with any possible future product if Earfun ever decides to collaborate with me again.