Everything changes... (Patreon)

I got the Nubert nuPro X3000 RC some days ago and immediately shot a video after unboxing, but instead of posting it immediately I decided to analyse the sound of these speakers a bit more in detail and set them up in a perfect acoustic triangle sitting about 130cm away from them with both speakers placed 130cm apart in order to perform some measurements, recordings etc and discovered some interesting facts which I would like to share with you.

I recorded the speakers with most my binaural microphone types I am owning in order to be able to match the responses for all of them and get comparable results for all of them in terms of correctness (to some realistic degree of course).

So far I only concentrated on my main microphones which I was using recently most. Those are closed type, thus they block my ear canal when I wear them, I cannot judge the recorded sound that well with them but they have a pretty low amount of hiss and offer mostly consistent results with repeated insertion.

The problems with my sound samples so far was, that they were not "neutral", in reality the recordings appeared too bright when listenend over a flat speaker, but I equalized them to sound correct over a "flat" equalized headphone compensating for the brightness, thus over the headphone it sounded correct again (but in reality the headphone was too dark to be neutral) and this is where the problem starts from the beginning: I am still unsure how "neutral" for a headphone really should sound like and was trying too approach this entire matter with my current speaker setup again, unlike last summer where I tried to use diffuse wind noises around me to come up with both a compensation for my recordings and also derive a headphone target from that. Unfortunately I don't own those microphones anymore which all my "research" was based on, but thank godness the work I did back then did also come in handy this time as well and helped me to speed up some steps.

With the speakers set up properly I should be able to derive a more precise compensation for my binaural recordings which should offer correct results for sound coming frontally from around 30 degrees to the left and right horizontally. Thus a signal like this recorded with my ears should correspond to that of a measuring microphone placed at the same position (to some realistic degree again). Sound coming from other angles (above, below, behind, to the sides etc) will get the frequency response altered according to my HRTFs, thus only the front direction should also sound correctly played back over speakers, other directions should still sound correct when played back over a headphone as the human ear will perceive the changes on frequency response as direction cues. I am still not sure how well my own HRTFs go along with others’ ears and prepared some sound samples for that task to get some feedback from you.

To demonstrate you better what I am talking about, I would like to show you some measurements. Below you can see the X3000 measured in my room with the measuring microphone waved around the area where my head would usually be. Overlaid in red and blue you can see the X3000 again but this time measured with my binaural microphones inside my ears and equalized for left and right separately to match the originally measured frequency as close as possible. Equalising for left and right ear should improve imaging and realism of space considerably compared to before where I just took an average for left and right ear:

If smoothened more, you can see that I really tried to stay as close to the original measurement as possible:

My new compensation averaged for left and right ear in green and compared to my old compensation in purple:

But the final step is to make this recording which in fact should be "flat" if listened over a flat system to also sound correct over headphones. As usual I used the HD600 to play back this recording and equalising this result to match the original measurement as close as possible. And here is where I am currently fighting: you can see that played over the equalized headphone the signal gets some bass bosst, with some additional dips and errors in mids and upper mids:

I could either compensate for these errors by reducing bass amount for the headphone, or by reducing bass within the recording, which then wouldn't be flat anymore. But ultimately we want the played back spectrum over the headphone to sound exactly the same as the spectrum heard live from the speakers. If my recording sounds correct over the headphone, then finally the headphone should also be "flat" and pink noise played over the headphone should sound similar to pink noise played back over speakers.

The final result was not that far off already before with my previous samples. Thus my old recordings played back over my previous EQ for the HD600 gave a similar final result like you can see in the measurement in green below, although this time I managed to improve the result even more and now at least my recording can be regarded as "neutral", while before the recording was too bright, and the headphone too dark as a result:

What does this mean? A "correct" headphone target whatever that means, is definitely quite a bit brighter than what most of you but also me are probably used to. I am still in the process of optimising everything. I tried correcting for the differences between measured response from the headphone to that of the speakers but it resulted in a wrong sounding timbre. Therefore I want to record more samples for the headphone by changing both microphone placement and headphone placement every time to get as much variation as possible to get a better average for the final EQ. I am currently listening with it and you can get used to it pretty quickly as the amount of detail you are hearing is sheer amazing. At the same time I want to fine tune it better by comparing different signals played back over the heapdhone or the speakers and they should still resemble each other. The only thing  I noticed is that the analysed spectrum from either pink noise or music will appear again different in the bass area. With music I measure a considerable bass boost in my recordings as opposed to pink noise, I may do again an average from these 2 as well to have as little errors as possible.

In the end this final target should allow for perfectly "neutral" monitoring of not only my own recordings, but you should get a more correct music reproduction over headphones in general. Besides my sound samples should contain less errors than before and should more or less represent “neutral” for sound being emitted from front of me.

I prepared some sound sample to demonstrate you the differences between my new and old compensation. Sorry about the background noises as my wife was walking around and cooking, but I uploaded a short binaural recording of the X3000 playing pink noise and a short excerpt of Analog from Soulpersona in both my old and new compensation mixing in the original signals for comparison. Thus you will start hearing the speakers switching to real pink noise some seconds later and so forth. Youu should notice that the new compensation should not only sound smoother and more natural overall but also closer to original pink noise and to the original recording of "Analog" at the same time it appears darker compared to my old compensation, this is now corrected within the headphone EQ. To play back these recordings correctly, you need to use this new EQ for the HD600 which is attached below. It is still pretty rough with lots of correction filters. Only when I manage to fine tune all errors as good possible, I will try to reduce the filter amount to 10 again to achieve the same original response as close as possible while still sounding “correct” and finally publish a new EQ. I would welcome those with a HD600 to also try out the other audio samples I did. You need use the attached EqualizerAPO file and you should hear the samples more or less how they sound in reality, minus some averaging and smoothing errors etc.

Let me know how you perceive these sound samples, and if the positioning of the plastic bag is rendered realistically for you as well. As we are still dealing with my own HRTFs captured within these recordings, I still wonder how well those can be adopted by someone else. For me the improvement in realism as opposed to my previous compensation for my recordings with both EQ for the headphone is significant. The plastic bag and water bin sound just as if I hearing them in front of me. 

This does not mean that the achieved target will or should sound great, at least it should allow for an as correct as possible reproduction chain. Please let me know if I should already post my upcoming videos with this new compensation!