Fast and responsive transducers are obviously good for headphones regardless of the use case. Same goes for lack of ringing, resonance and smooth response. But are these specific to binaural rendering?
Id say yes, with binaural rendering objective performance is most important, as these traits aren't necessarily even desired for normal headphone listening. Some like grados boosted distorted treble, some like beyerdynamics ringing/resonance, some like vmoda boomy out of phase bass, but any of those headphones provide a less than desirable experience when using hrtf convolution and especially prir measured from loudspeakers.
Sure it would be easy to think that there is more information that the transducers need to reproduce when there are multiple channels being played through them but this of course depends on the source material. In movies the center channel doesn't really contain information that wouldn't be in the front left and front right channels if it didn't exist and side and rear channels mostly contain ambient sound which are not rich in detail.
There is definitely a substantial amount of additional information even when just listening to stereo music, the binaural rendering functions similar to crossfeed adding left channel information to the right channel, in addition to time and domain processing. And then adding the reverb to every single sample, imagine the size of a lossless 7.1 mix with a 14mb/s bitrate.
Angled pads, open back enclosure, pinna interaction and planar waveform might be things that have an effect on localization with binaural rendering although I would be interested to read some science behind this.
I dont think this is true for localization as much as externalization. Ive tried a fairly wide range of headphones with OOYH, from in ears to the EL8s and the staging ability of the headphones has a large effect on externalization. rtings does soundstage meaurements that can actually be rather interesting and insightful and correlates well with my experiences with virtual speaker locations perceived as further away. Localization would be more tied into imaging, for headphones, this would be driver performance(seen in 300hz square waves and impulse responses) and smooth treble.
Heres a good example of smoothed treble, the dekoni lcd velour pads on a pair of audeze lcd2c vs the stock pleather pads.
In my experience very convincing localization and externalization can be achieved with affordable headphones like HD 518. Convincing to a point where I often find myself checking if I'm actually listening to headphones or the speakers in front of me because I cannot tell from the sound. However I have to confess that I tested these headphones with acoustically personalized HRIR measured with ear canal blocking microphones and I also do headphone frequency response compensation with the same mics in place.
With personal measurements yes you should get good externalization, thats how the sxfi was demod originally wasnt it? Aside from the headphones, Ive gotten the best externalization using the prir presets with OOYH compared to anything else, even without being personalized measurements, and localization in a 7.1 mix is excellent, the most I've found I had to adjust is channel balance, which could either be due to my head or the measured head.
Headphone compensation might be the important thing here because Super X-Fi doesn't have it and therefore it cannot compensate for variations in frequency response between individual headphone units. For these kind of use cases headphones with tight manufacturing tolerances and good consistency are clearly preferred. Even more important might be transducer matching because if the left and right side transducers are very different and there isn't compensation for this then the localization is not going to work as well.
Yes being able to perfectly compensate each channel individually would be ideal, but simply flattening out the uncompensated frequency response of the headphones and even equalizing any very noticeable colorations when say playing back a sweep with sxfi or ooyh running, will provide a much more enjoyable experience.
good for headphones regardless of the use case.
Back to this, as you said the attributes are beneficial for headphones regardless of the use case, this is especially true for hardware and why sxfi really needs to provide a software only option. The differences is externalization, naturalness(especially with a Realiser or OOYHs measured presets), and fidelity are astounding as better equipment is able to better render the increased detail, and that detail being the key to the externalization i.e. room reflections. Using a usb dongle or the motherboards builtin audio makes these binaural renderers sound like a neat effect, theyre not able to render the detail, slow circuits feedback delay etc and the information required to truly hear whats being rendered is lost forever. Using a unit like the audio gd nfb11, with it's discrete dc coupled class a no feedback current signal amplifiers, ultra regulated, ultra fast, short signal path, turns that neat effect into a large and full image. Adding conditioning devices like a topaz isolation transformer for ac noise, and an uptone iso regen for usb isolation and signal integrity is much more noticeable when using OOYH, taking that large full image, which can actually be fatiguing with certain material do the massive amount of extra information, and yielding ultra clean ultra detailed non fatiguing sound. Take for instance the dc power cable between the iso regen and its power supply, the wide current draw of digital signals creates a voltage across the inductance of the cable which creates noise, this noise can't be dealt with by the devices regulators, and you alleviate it by using low inductance quadrapole(star quad cabling). Thats just 1 layer of noise that propgates through a system, common/traverse ac line noise, noise from smps leakage currents, phy noise, etc. And cramming a whole bunch of feedback into a transistor amplifier just distorts transients for the sake of great distortion, signal to noise, and frequency response measurements, yay flat to 80khz but etched and dull sounding. Moving up from there to something like audio gds d27 with full balanced drive takes that large and full clean image and makes it incredibly lifelike and natural sounding. Its shocking how much more speakerlike even OOYHs genelec preset sounds on the nfb7 compared tocthe nfb11. I'm sure I've lost 99% of you at this point but I assure you, no foolies.
ε-(´・｀) ﾌ
All my comparisons are done with complex movie scenes like the ready player one race, fast complex death and black metal like anata and nile, or high quality recordings of solo instruments or 5.1 orchestral peices. This kind of material played through OOYH makes differences in components easily identifiable, differences that I dont think I'd be as likely to hear with normal stereo headphone listening.
I don't see how HeSuVi would be a problem. It uses EqualizerAPO under the hood and EqualizerAPO's convolution implementation is as good as any. If your comment was really aimed towards the HRIRs that ship with HeSuVi then I agree. Not having any form of personalization is quite the blocker for plausible localization. I have found only one HRIR which isn't personalized for me and can do decent frontal localization. But HeSuVi with acoustically personalized HRIR is one hell of a thing.
Im not sure what it is, testing the ooyh genelec preset on hesuvi vs using ooyh sounds terrible, my first guess would be the hrir, since its measured from ooyhs output and not the actual original prir wave file. I just wanted to do my emoticon joke... but truthfully how you couldn't hear a difference between the hd518 and he400i is beyond me and leads me to believe you're not experiencing the full potential of binaural rendering.
I disagree that personalization is mandatory for localization. I think moreso for correcting colorations of the hrtf. And modeled hrtf convolution yields just fine localization, its just not that externalized, it takes at a minimum some kind of first reflection finite impulse response to get the virtualization out of your head, and personal room impulse responses to get it out into the room with any sort of fidelity(waves nx 100% room ambience sounds huge but not so great).
So hey not trying to be argumentative, just providing a perspective on all this that's maybe not so common.
(Your mileage may vary. Offer may be limited in certain territories and time periods...)
