HRTF and soundcards

[K.I.L.E.R]

I'm spinning off a totally different discussion in this thread all about HRTF.
Start:

Only if they support HRTF.

I knew a guy in my class 2 years ago he bought $1000+ headphones which do HRTF for second hand at $400.

Hey, I had one of those too - great card It's been said before, and it will be said again, but A3D was soooo much better than any version of EAX I've heard. It also suited my prefered sound method, which was headphones (which when done correctly always sound better than speakers, IMO).

How do soundcards emulate HRTF and how is it different from hardware doing it?
How would a soundcard and a speaker/headphone work toegether to achieve HRTF?
Do they need to work teogether to achieve a fully realistic HRTF effect?
Why isn't HRTF being supported in more games?

 

[_xxx_]

Why isn't HRTF being supported in more games?

For the same reason we don't do raytracing on gfx-HW

 

[K.I.L.E.R]

We support mutiple sound sources on multiple speakers.
Isn't that more intensive?
Why is there HRTF hardware?
Why do some games have HRTF?
NWN has it.

For the same reason we don't do raytracing on gfx-HW

 

[KimB]

Well, the head-related transfer function is a method of encoding the 3D position of the source of a sound wave so that once the sound wave has passed through our ears, it sounds to us as if it has come from a very specific direction. With a good HRTF, for instance, you can hear a sound coming from behind you while only using a two-speaker setup.

Now, one of the keys with HRTF's is that all that you need to produce one is a direction from which the sound arrives. So you don't need game support at all. Presumably, all that you need is for the game to support a 3D sound API (DirectSound3D, EAX, or, once upon a time, A3D), which will in turn transfer this directionality information to the sound card's driver for the sound card to use the HRTF to calculate the proper modifications to the sound wave.

The reason why you have some games supporting HRTF's explicitly is because since the death of Aureal, HRTF support in sound cards has been abysmal. Games shouldn't need to support HRTF's at all. But some do because if they didn't, then most users would never get to hear the 3D positional audio.

The really sad thing is, though, that Aureal's A3D 2.0 wasn't just about good 3D positional audio, either. It also supported raytracing of the sound waves through the game world (just one bounce, but still very nice). The resulting effect was that if you hear a sound in a hallway adjacent to the room you're in, you hear the sound coming from the door instead of through the wall, in addition to it being somewhat muffled.

When combined with the excellent 3D positional audio from the HRTF's the cards supported, the above made for a really amazing playing experience back in the day. It's utterly unbelievable to me that this technology seems to have died utterly.

P.S.
An HRTF calculated properly in a sound card is going to be vastly superior to one calculated on the fly from headphones. The headphones don't get correct position information: they can only infer it from the audio output.

Edit:
Oh, yeah, unless the headphones support Dolby Digital decoding with an S/PDIF connection. I think that would give them the 3D positional information. But since no modern hardware uses Dolby Digital...

 

[swaaye]

The problem is most humans don't care that much about game audio, or audio at all really. Most people don't really perceive the complexity of their real world hearing, let alone realize the potential for reproducing it in games.

This is plainly obvious by how few people notice the abysmal audio quality in onboard audio on mobos. Hell some of the mobos out there might as well be specced as producing mono sound with midrange only lol. People just don't know that there is potential for so much more. Audio is a very non-visual experience and I think people are far more visually oriented on the whole sensory importance angle. It's like how right now I'm sitting here thinking of listening to tunes but am not quite motivated enough to do it, but I know inside that once I put my phones on I will quickly be at the point where I can't imagine not having them on.

Now, if you sit someone down in front of Half Life with A3D 2.0 running on a Vortex 2 and put them in a firefight, you will probably see the light come into their eyes as they hear the result. Assuming they can get past the old graphics, which most people whine about immediately unless the game's a fav pasttime.

 

[KimB]

Yeah, I think the problem is just that people don't notice a problem. They can hear the game just fine. And unless their rig is seriously cheap, there probably won't be any nasty hissing or anything like that, either.

But I think an enterprising company could make it work well today. It would require a combination of a good product and good marketting, but I think it could possibly be done. Or hell, nVidia could cause a revival easily if they just implemented it with their next generation of nForce products. But I think they feel that the R&D money for such a product would be largely wasted.

 

[_xxx_]

The real problem is that HRTF, if properly done, requires tons of preocessing power and will let your framerates sink, sink, sink...

Even that rudimentary wave-tracing in Vortex2 would halve the fps, now imagine what would happen with a more intensive algorithm.

 

[KimB]

Even that rudimentary wave-tracing in Vortex2 would halve the fps, now imagine what would happen with a more intensive algorithm.

That was a long time ago. The performance hit was fairly significant back in 2000 (or somewhere thereabouts). What do you think it'd be today for the same wave tracing algorithm? That's right: next to nothing (and yes, though today's games have more geometry, there's no need for the sound card to know about it all: you can always produce low-geometry versions of levels).

Anyway, as for the HRTF, yes, it can require quite a bit of processing power. But that's what dedicated hardware is for, isn't it? We've had dedicated sound cards for years, why shouldn't we expect good HRTF's?

 

[swaaye]

I've always sort of seen the Vortex chips as dedicated 3d audio processors for gaming. The EMU10K1, on the other hand, was really a high-end MIDI chip (at the time) which could use its effects engine on non-MIDI I/O too. EAX was just the reverb/chorus engine running on your digital audio. So, they were really very different. Neither was bad I'd say, but Live! wasn't really tailored for 3D audio like Aureal's chips.

The newer cards from Creative are more capable of interesting 3D. CMSS2 uses something called binaural virtualization in headphone mode to give you a different sound. It's not necessarily better though, especially for music. X-Fi is supposedly quite nice in headphone mode. At least that's what most reviewers rave about. Who knows if these reviewers have much background to compare to.

When Vortex 2 was hot, back in 1999-00, we were running PII, PIII, and K6-2/3. Considering how what they were doing back then was generally superior to almost all cards today, except potentially the Audigy2/X-Fi, I'm sure the performance hit would be absolutely trivial if they stuck with similar tech. We are enjoying generally far higher framerates today anyway. I'd gladly lose some frames for some new audio tech. But I think Aureal proved that the market isn't really there, and so did the excellent nForce APU.

DirectX has had some HRTF algorithms in it since DX7, btw. If any of you've played Painkiller, you'd have seen it in the options. That game used the nifty RAD Game Tools' Miles Sound System and supported damn near every audio standard out there.
http://www.radgametools.com/msssdk.htm

Digit-Life whipped up a wicked article on game audio a year ago or so too.
http://www.digit-life.com/articles2/sound-technology/index.html

 

[Anarchist4000]

One other thing to keep in mind is that with the introduction of dual core CPUs audio processing is something that dual cores might come in really handy for, even when using a software based setup.

 

[K.I.L.E.R]

Can someone tell me if I should buy a $1000 sound card to hear good HRTF IN ALL GAMES from my stereo speaker system?

 

[KimB]

Can someone tell me if I should buy a $1000 sound card to hear good HRTF IN ALL GAMES from my stereo speaker system?

I would certainly not plop that much money down without some assurances as to how well it works, and whether or not the difference is worth it to you.

 

[_xxx_]

Can someone tell me if I should buy a $1000 sound card to hear good HRTF IN ALL GAMES from my stereo speaker system?

I'm not really that familiar with that, but AFAIK there is no card out there which will do HRTF in every game/app. I may be wrong though.

 

[KimB]

I'm not really that familiar with that, but AFAIK there is no card out there which will do HRTF in every game/app. I may be wrong though.

Yeah. To do it, you'd need to reconstruct sound positioning data from the sound sent to each speaker. And even then for most games all you'll get is stereo, which isn't what HRTF is all about: you want front/back, up/down info to get the most out of HRTF.

 

[K.I.L.E.R]

I'm much more knowledgabe in physics than a Phd because I have 12000 Phdz in craptalking but I would like to know why the obvious hasn't been done?

3D Game = Sound is in position (x, y) in relation to playa.
Sound card = do phytsics in relation to sound + playa
Speakers = make sound

So I assume as in 3D sound physics requires a lot of half assed broken down physics formulas called "approximations" to get decent performance?

 

[Zaphod]

3D Game = Sound is in position (x, y) in relation to playa.

X, Y and Z. Just X and Y give you only 2D sound. Mono = point source, Stereo = 1D, Surruound = 2D. Throwing more speakers at a setup will obviously improve 2D positioning to a degree, and if you have a 7.1 system there may not be much need for extremely advanced algorithms to prevent a panning sound from making all too distinct jumps between speakers.

The trouble is that most people never set up their systems properly, have a listening position far from the sweet spot, and crosstalk between each speaker (the sound meant for one ear also being being heard by the 'wrong' ear, and the speakers inevitably interfering with each other due to positioning when you add more of them) tend to cause any 3D cues, in particular with regards to elevation (hence the third dimension), to be lost.

 

[K.I.L.E.R]

Sorry I was eating toast with bread and chicken.

I don't want more than 2 speakers. I want perfect audio with 2 speakers.
My speakers are set up perfectly. There is no crosstalk. They are fairly distant, most important of all they are aimed at my ears and have books under them acting as adjustable stands.

X, Y and Z. Just X and Y give you only 2D sound. Mono = point source, Stereo = 1D, Surruound = 2D. Throwing more speakers at a setup will obviously improve 2D positioning to a degree, and if you have a 7.1 system there may not be much need for extremely advanced algorithms to prevent a panning sound from making all too distinct jumps between speakers.

The trouble is that most people never set up their systems properly, have a listening position far from the sweet spot, and crosstalk between each speaker (the sound meant for one ear also being being heard by the 'wrong' ear, and the speakers inevitably interfering with each other due to positioning when you add more of them) tend to cause any 3D cues, in particular with regards to elevation (hence the third dimension), to be lost.

 

[Moloch]

For perfect 2 channel audio you need 3 speakers
(sub).

 

[K.I.L.E.R]

Isn't that 2.1?

2 + 1 = 2.1?
1 + 1 = 10
1 + 1 = 0x02
1 + 1 = window

I have that.
I have a sub woofer.
A sub is liek a subway sandwitch.

Also do I need to get Creative's latest soundcard to get the best possible 2 speaker HRTF audio?

For perfect 2 channel audio you need 3 speakers
(sub).

 

[BRiT]

I don't want more than 2 speakers. I want perfect audio with 2 speakers.

That would have been A3D version 3 -- genuine audio wave-tracing. Even A3D rev 2 blows away the reverb Creative calls EAX. *sigh* If only Creative would have done something with all the cool tech they scavanged from Aureal. They took some steps towards that with EAX 4, but it still seems to be too far away from actual positional sound Aureal was getting at.