The various methods of Virtual Surround (and how shit they are)

With Out of Your Head, you should equalize your headphones to have a generally flat uncompensated response. This is done during setup with the Realisers but you have to do it manually with OOYH.

Second, use more responsive low distortion drivers like planar magnetics/electrostats, and stick to headphones with high pinna interaction(i.e. hifiman he400i/audeze el8 with dekoni lcd velour pads over stax sr207, the Realisers are often demo'd with the HD800).

Your dac and amp make a big difference in rendering the binaural stereo signal, dont expect your pc motherboard to sound good, between the binaural stereo and the measured room reflections inferior equipment simply fails to render all the information.

I use OOYH 100% of the time for stereo music and the fidelity is astounding when you put forth some effort to equalize your headphones and use equipment better suited for the complex audio signal.

The most I find myself doing when correcting for hrtf prir that isn't my own measurement is channel balance and additional equalization in certain areas only if its very noticeable during frequency sweeps.

 

I know this was a year ago, but would running something like a vali 2 into the GSX 1000 forego the need for these EQ compensations?


Also, has anyone tried Super XFI?

 

Virtual surround has many uses.

If your input to your Virtual Surround Processor (VSP) is stereo, all you will get is phase-erroneous mangled garbage, imho. It may sound perceptually more spacious and have lots of added echo / reverb added to it, but there will be no realism or accuracy.

However, if your VSP can input discrete, multichannel or even object-based 5.1 - X.Y sound signals (unprocesses, not yet downmixed to stereo), then you can potentially get more "lifelike" and "accurate" virtual presentations on your stereo headphones, IFF the VSP does it's job properly.

Unfortunately, most VSPs just add echo and reverb (approximate/simulate room reflections). This is cheap/fast to do, but also dumb. Let's say your original surround X.Y discrete signal portrays animal sounds in open air jungle recorded in real nature. Having added / artificial room reflections/echo/reverb on top of this, just sounds artificial and stupid.

Then there are better VSPs that try to approximate HRTF phase/FR changes by applying different time delay and frequency changes to different incoming discrete signals, before downmixing them to the stereo virtualized sound. Unfortunately, I know of NOT a single commercial one that does without added artificial room echo/reverb simulation (For example DTS Headphone X, Dolby Headphone and Creative Super X-Fi all add artificial "room acoustics" to your discrete signal during the VSP processing).

What is also an issue that almost all HRTF based commercial implementations use a generic, one-size fits all HRT function to apply to all users. Perhaps only Creative SX-Fi currently allows taking a photo of your outer ear and uses this photo to "customize" the HRTF used more towards your anatomy.

Then there is HeSuVi. which is free and IFF you want to set it up properly, a PITA to do properly, but can be the closest to what we can get now (as long as we don't have object-based audio output and real Object based audio input VSPs AND personalizable HRTFs).

To set up HeSuVi for NON-echoic & personalized HRTF, you need to record your own Impulse HRIR measurements with binaural microphones (ear canal blocking microphones) using software Impulcifier in your own ears, using the surround channel speaker setup you want (if you want to virtualize 7.1 speakers, you need to have a proper real 7.1 loudspeaker setup to do this, in addition to good quality binaural mics in your ears). Further, if you want this without the artificial echo/reverberation, you need to perform this in as acoustically dry / dead environment as possible (quite close to, but not totally acoustic anechoic chamber. How many of us have access to something like this? Not many).

Once you have have your own HRTF response measured with Impulcifier, you can use this file in HeSuVi to have accurate (as accurate as currently possible) of VSP of multichannel discrete audio signals down to virtualized 3D stereo using your own HRTFs.

The benefits of Impulcifier DIY + HeSuVi:
- You can pick your loudspeakers for audio reproduction (7 full frequency and stereo sub-woofers, if you have access to such gear) that you will simulate
- Personalized to your head/torso/ear(lobes) (even if not ear canal)
- You can get rid of some/all of artificial room reflections/echo (the close your recording environment is to an anechoic chamber)
- processing latency is very low (in milliseconds. The worst of hardware/software based 3D VSPs for gaming add an additional latency of UP TO 200 ms. Try competitive gaming with that. That's a 1/5th of a second of delay). That's a delay of 50 frames or more on competitive high Hz gaming displays.

Of course, doing this properly and paying for all the gear or renting it, will cost you several hundreds if not thousands of dollars/euros.

But it will be as accurate as possible, no artificial reflections, tuned to your outer ears and have minimal delay.

Will it do anything useful to incoming stereo signals (say you play back high-res stereo music) : of course not. It will only mess it up. This works for incoming multichannel discrete audio signals, in order to try and reproduce the subjective experience of that using stereo headphones.

Now, as the last option Impulcifier+HeSuVi+binaural Mics+near anechoic chamber is out of the reach of 99.999999999999%, then what is the best?

It depends on what you want? My current takes are, not having tested them all:

- Best sound quality/truthfulness/honesty : HeSuVi with minimal processing (try any of the algos in the pre-built package off Sourceforge - the less echo/reverb the better)
- Best 3D audio signal positioning / accuracy / determining incoming direction (esp. useful for competitive 3D fast paced FPS games) - my current pick Audeze Moebius or Creative Super X-Fi (if you can get latter to personalize properly your ear transfer functions). Super X-Fi creates much more delay than Mobius NX (even in wired mode)
- Least amount of artifacts : hard to pick, they all suck bad, I'd take HeSuVi with generic (non-personalized) algos and play around with those. For this you really need to do your own recordings of HRIR
- Widest compatibility with most headphones? Creative SoundblasterX X7 or G6, as it has digital output (to feed into your external dac/amp) and has enough juice with its built in headphone amp to drive many headphones.

 

Picked up a Creative Labs SXFI X4 the other day.

Being the latest SXFI processor hopefully it can do it without adding a ton of delay the way the cans with it built-in do.

If nothing else, it has a neat built-in mic controller for games with an always open mic, like Forza Horizon, and I'll just run Dolby which is the least offensive of the software packages, and the TOSLINK out on the recent creative stuff is pretty solid for piping into the Yggdrasil.