If you are familiar with head-fi, chances are you have already come across those words many time: HRTF (Head Related Transfer Function) or put in a different way, we all have different ears.
It seems like an indestructible axiom on those boards, and often used to explain two radically different opinions of the same gear, for example headphones. It is a powerfull tool to remove tensions in heated discussions. But what bothered me with those claims is that I have actually never seen any analysis to back it up, nor quantify its importance. Is HRTF a question of minute differences (less than 1dB) in the way we hear things or rather large and significant differences that could indeed explain why two people can have radically opposed opinions of a single headphone.
That is what I have tried to find out.
I was very much interested in this aspect of our perception of sounds, as I was wondering if true neutrality could indeed exist in a headphone and be somewhat universal, beeing myself a firm defenser of true neutrality in a playback system.
One very interesting and preleminary article can be find here, in the Stereophile magazine, Between The Ears: the art and science of measuring headphones. I learned there that Free Field calibrated headphones and Diffuse Field calibrated headphone would have inherently different frequency response, as well as many other valuable information.
But what got my attention was concretely the first footnote of the Stereophile article, as for the first time, I could actually see a reference to a serious analysis and study regarding the HRTF disparities between individuals. It is a paper from the AES, and not free, so I will not divulgate its content here. However, and if allowed, I just would like to show 1 graph extracted from this paper I bought and is treating HRTF disparities...
What is represented is measurments for 40 individual HRTF in Free Field conditions (left side), in a controlled environment (speaker, not headphone, but still valid I believe for our needs). There is also a similar graph for Diffuse Field response, but the conclusions are the same. Pay attention to the half bottom of those graphs, showing HRTF with open ear canal. The dark grey line on the right side is the average for those 40 measurments available, and the light grey area is indicating one standard deviation. You can actually see how perception between individual can vary easily in 10dB (standard deviation at 10khz for example corresponding IIRC to only 65% of the population analyzed, or 27 individual HRTF), and even more for those than are not into this standard deviation (you can actually spot difference of 20dB or more). Impressive stuff really, and an eye opener for me definitely.
Additionnally, you can observe that those differences are appearing only in the high frequencies range (above 2khz)... Below 2khz, we all hear "more or less" alike, within a confortable tiny range of 1 or 2 dB. That means that different perception in the bass response of a headphone is more linked to a matter of personal preferences than a real difference in HRTF. Very interesting. In that paper, they explain this because low frequencies are almost not altered by the human anatomy and are able to pass through our body with very little changes. The same is not true for higher frequencies, where our body can indeed play a pass band filter role, and where differences in ears anatomy can be relevant. Makes plenty of sense to me.
So, does true neutrality really exist and is it universal ? I believe the answer is yes, but unfortunaltely not for everyone. You can design a headphone to be as neutral as posible (note that the frequency response is far from a flat line), but it will only sound so to people with average HRTF response. For those persons who present significant deviations from the average HRTF, neutrality will be much more difficult to simulate.
However, if a significant majority of the community find a headphone neutral, then it should - by definition - be considered as neutral. This does not mean that in some isolated cases, impressions of single individuals would not meet the overall consensus. Indeed, one should try to learn on head-fi if they usually fit with the consensus on a headphone or if they are rather a minority voice, and always keep this factor in mind when reading other people opinions.
I hope this reading was constructive and interesting to you. It certainly was to me. Thank you.
Edited by shamu144 - 10/15/10 at 2:20pm
