3d Audio with Headphones

[Irongrip]

http://en.wikipedia.org/wiki/Binaural_recording

Has valve considered implementing Virtual Binaural Support for their games?

[Seth.Sekhmet]

Lol, thanks for the article link, but I don't believe it would be such a noticeable feature. Do you have audio samples ?

[Irongrip]

They are linked from the article.
And believe me a lot of immersion comes from where you can hear sounds.
Distance to them would allow you to imagine the scene more deeply.
Since all sounds are "directed" you already have a system that could be modified for when you're using headphones...

[WildCat_ZA]

This is a feature that I think should be given much more serious attention by game developers. For a long time, I've wanted to see this sort of feature included in games. I'm not interested in buying a surround sound speaker set because I don't have space for it in my room and it would annoy my family. A quality pair of headphones is the answer - I just wish they could produce realistic surround sound.

There's a feature on my X-Fi sound-card called CMSS-3D, which is supposed to create this sort of effect but my experience with it is that it's not that convincing. When I enable CMSS-3D and set Team Fortress 2 to use 7.1 surround sound, I can discern a difference between sounds coming from behind and from in front but it's not particularly realistic. It works, but not very well.

If Valve could upgrade the sound processor in the Source engine to reproduce the effect of the head-related transfer function (http://en.wikipedia.org/wiki/Head-related_transfer_function) for people using headphones, I'd be a very happy man. Alternatively, perhaps a company like Creative Labs could create a new sound card that could off-load all sound computations onto its own processor and expose an API for developers to input the raw sound data and positional information for it to process into binaural output.

Either way, I'd really like to see binaural sound for headphone users in games.

[Darkimmortal]

I second this request. :)

[Irongrip]

You should take a look at this:
http://iem.at/Members/noisternig/bin_ambi
GPLd, too!

[Geister]

I don't think any of you realized this is used for music and movies, where the spectator is static. In a game, the player is dynamic, the direction of the sound constantly changes, therefore the binaural-recorded sound would not give a good effect, since it would assume that the sound is coming from the front, when it can come from any direction. Binaural effect is already achieved by most 3D sound processors.

However, sound effects and speech for cut-scenes could indeed be recorded that way, but as soon as the game would resume, it would have to revert back to 3D audio with normally-recorded sounds. As for head-related transfer function, it is available with EAX and has been for years.

[Irongrip]

You assume wrong, binaural sounds can indeed be recorded. BUT they can be simulated too. The game already has a system to track in 3d where the sound comes from. But it can also do much more, like simulate the binaural difference between the two headphone speakers. THUS givving you the illusion a sound is comeing from somewhere. AS you spin around to locate it, it could change it's position IN REAL TIME.

:D :)

[WildCat_ZA]

I don't think any of you realized this is used for music and movies, where the spectator is static. In a game, the player is dynamic, the direction of the sound constantly changes, therefore the binaural-recorded sound would not give a good effect...

The request is not for game developers to record their sound effects using dummy head recording (http://en.wikipedia.org/wiki/Dummy_head_recording). Rather, the request is that game developers program their sound engines to digitally reproduce the HRTF for headphone users in real time.

[Geister]

The request is not for game developers to record their sound effects using dummy head recording (http://en.wikipedia.org/wiki/Dummy_head_recording). Rather, the request is that game developers program their sound engines to digitally reproduce the HRTF for headphone users in real time.
Oh, I'm sorry, I guess I just misunderstood. Sometimes people find out about a "new" technology (most likely been around for years) and they suggest something that has been available for ages, or that would not apply to gaming.

I know EAX/A3D has been using this for years. I also found out EAX causes some problems with the Source engine audio (it uses Miles). I can't seem to find any info on Miles as to what it implements, but I wouldn't be surprised if it did implement it. It's pretty advanced middleware and has been around for decades. You'd need to talk to RAD Game Tools if you want this kind of feature implemented in Source games... or convince Valve to change the sound system in the Source engine...

But it can also do much more, like simulate the binaural difference between the two headphone speakers.
I know a lot of 3D sound engines already do it, and most likely Miles does it. Just set the proper speaker configuration and it should work.

[Boff]

this tech has been pottering around since the late nineties, when I was at uni, and quake just came out,
but no one has ever used it (wisely/properly) sadly


thanks for the resurgance of interest, as I've always wanted this!!!!

[Irongrip]

I have the sneaking suspicion that if you have the proper hardware it might do it on it's own. Alas i'm using a realtek 6 channels standard mobo soundcard. Setting HL2/TF2 to use headphones and doing the same in windows' control panel to the same does not make the game use binaural sounds, only stereo and the sounds appear to distinguish only from left-right :[

[Nickname P]

This same suggestion was posted by me several years ago, and nothing happend till then :(
Also this audio sample, too.
http://www.metacafe.com/watch/821855/virtual_barber_shop_use_headphones/

[Boff]

that virtual hair cut was stunning!!!

[Boff]

seriously what happened to binaural?

it was all the rage 10 years back.

but died.

and literally been looking around (for about 3 minutes) and especially for this cetera algorithm.

no official site or anything just references to it on forums and what not, and old references at that.
one even dates BACK to 1999.

with valve doing wondered with multi-core, especially offloading sound to a core,
you'd think they couldn't do some of this stuff on the fly.

[Geister]

Maybe there's a good reason why it is nowhere to be found anymore. It's nothing a 3D sound system can't simulate. It's all about the position of the sound. Binaural recording works only if the subject listening to the sounds is static, that is not moving, and if you don't like the effect it gives, you have to record it all over again. With 3D sound, you can just "move the sound around" if you don't like the effect it gives, no need to record again. It also allows the subject to be moving in the world, rather than have the world move around the subject.

[slayer20]

Ok that was just creepy.

"Hi there. How you doing today? Good."

[NeryK]

Slightly off-topic, another option is to use surround "5.1" headsets (for instance Tritton AXPRO in the U.S. or Sharkoon X-tatic in Europe). You get a fairly nice boost to your game immersion (particularly in games like L4D where positional audio is well executed). Of course it is not perfect, but if you have the money, I find it is worth it.

Word of advice though, avoid USB headsets with integrated sound chip no matter the cost. Analog or optical is the way to go.

[Boff]

i had som zalmans a couple of years back, exceedlingly tinny noise,
worked well with doom3,
but not with source games,

but then again that could be because of my Realtek onboard chipset (AC97 i think)

i don't want to waste any more hardware cash, if the effect can be done, and done more effectively, via software.

i will front investment money into this field if valve wishes :) so they can generate a tech demo :)


It's odd, even the depeche mode instrumental songs snowblow, painkiller (maybe it was remix of that), tried to have various binarual effects that worked well in headphones. And this is back in 1996, right on the crest of the binaural wave when it hit, and probably achieved with a lot of labour in the studio to achieve binaural effect.
but never again did they do such a thing

[WildCat_ZA]

Maybe there's a good reason why it is nowhere to be found anymore. It's nothing a 3D sound system can't simulate. It's all about the position of the sound. Binaural recording works only if the subject listening to the sounds is static, that is not moving, and if you don't like the effect it gives, you have to record it all over again. With 3D sound, you can just "move the sound around" if you don't like the effect it gives, no need to record again. It also allows the subject to be moving in the world, rather than have the world move around the subject.

I think you're misunderstanding the concept slightly. If game developers were to implement a binaural effect in their games for headphone users, they would not need to make any changes to the way that they record their sounds. It would merely be an extension of existing 3D sound processing. For interactive games, the binaural effect would be produced in real-time by the game engine or a sound card by applying modifications to sound effects based on their position within the game world relative to your character's head. The HRTF information is not imprinted in the sound effect that is distributed with the game - It's applied digitally in real-time while playing.

-----------------------------------------------------------------

It's currently possible to get a binaural effect in a game like Team Fortress 2. Here's how:
You need an X-Fi sound card (or other sound card with CMSS-3D or equivalent)
You need to enable CMSS-3D in the sound drivers before you play (and disable it when you're done playing because it messes up music).
You have to set the audio output to 5.1 or 7.1 speakers in Team Fortress 2's audio options (even though you're using headphones)
This will produce a vague binaural effect but it makes the game sound rather 'tinny'. It would be much nicer if this sort of effect was native to the Source engine and could be switched on and off via the audio options menu.

[WildCat_ZA]

I just discovered that Unreal Tournament 3 sounds incredibly awesome with CMSS-3D enabled. Unreal Tournament 3 supports Open-AL so you simply enable that in the game's audio menu and enable CMSS-3D in the Creative drivers and you've got a binaural effect that's actually good.

The latest version of Unreal Tournament 3 has some new content including a map called "DarkMatch". This map has some funky, high-voltage coils going around the map that have electricity coursing through them. It sounds fantastic when the buzzing electricity goes by your head. The binaural effect also really helps with pin-pointing enemy locations.

The only issue is that CMSS-3D applies to all audio, including the music, which is not ideal. The music is a stereo track and sounds slightly strange. It would be better if this sort of thing was integrated into the game by the game developers. Then they could control what it applies to and make it sound even better.

[Irongrip]

I guess there is no way to patch that CMSS-3D feature into the drivers of a say, Realtek sound card?

[WildCat_ZA]

I guess there is no way to patch that CMSS-3D feature into the drivers of a say, Realtek sound card?

I wouldn't know but I'd be surprised if there was.

[Stryk]

http://en.wikipedia.org/wiki/Binaural_recording

Has valve considered implementing Virtual Binaural Support for their games?

This would be best.

But people can't let go of the thinking "more is better" and will cling to the idea multiple speaker headphones are better.

Aside from a base speaker, two high quality headphone speakers, one for is each ear with the proper processing is as good at it could get. But there's no convincing some.

[Geister]

Just remember that the more processing you put in the audio, the more delay you induce, the better the computers need to be not to lag. By adding software-computed HRTF, you just took a lot of processing power from the rest of the game, such as physics. Also, the effect is only noticeable with headphones, and since not everyone uses them, it never was really a priority for game developers. However, sound card manufacturers, such as Creative Labs, have integrated that in their card for a while, with EAX.

As for binaural output, it has always been part of 3D sound engines. If a sound is coming from the right, it will play loud in the right ear, and will be faint in the left ear, so you know where it is coming from.

I remember playing Call of Duty 4 with a USB headset and I could tell where all the enemies were just by listening to them, and even if they were above or below me. My sound card obviously did not support any special features (built-in with the headset). You simply have to listen carefully, but in fast-paced games, it was impossible to say the elevation, since the action was too fast to properly localize the sounds. Try getting a surround sound system and place it all around you. Trust me, you'll hear and even feel that helicopter pass right above your head.

http://en.wikipedia.org/wiki/Sound_localization#Binaural_cues
If you want HRTF, get a card that supports it, such as one with CMSS-3D, because software-wise, this is too demanding for the machine. As for binaural sounds, it already exists and existed for decades.

EDIT:
But people can't let go of the thinking "more is better" and will cling to the idea multiple speaker headphones are better.

Aside from a base speaker, two high quality headphone speakers, one for is each ear with the proper processing is as good at it could get. But there's no convincing some.
Headphones with just two speakers (left and right) are unable to properly give you the elevation of a sound, especially in a fast paced game. If you ever get to try a surround headset, you'll never want to go back.

[WildCat_ZA]

Headphones with just two speakers (left and right) are unable to properly give you the elevation of a sound, especially in a fast paced game. If you ever get to try a surround headset, you'll never want to go back.

It would be possible for a two-speaker headphone to give you the elevation of the sound if the game (or sound card) had a good HRTF implementation.

Sound that enters the ears from above sounds different to sound that enters the ears from below. Sound waves are changed as they interact with your head and ears before reaching your ear-drum. Provided that these changes could be simulated accurately through digital sound processing, you'd be able to discern elevation differences in sounds with just a stereo headset.

[Geister]

It would be possible for a two-speaker headphone to give you the elevation of the sound if the game (or sound card) had a good HRTF implementation.

Sound that enters the ears from above sounds different to sound that enters the ears from below. Sound waves are changed as they interact with your head and ears before reaching your ear-drum. Provided that these changes could be simulated accurately through digital sound processing, you'd be able to discern elevation differences in sounds with just a stereo headset.
But think about all the processing that takes. It just is not practical for a real-time simulation, especially with this many sounds playing at once.

Getting a sound card that supports it is the best solution for that. Or you can write a HRTF algorithm and implement it in a demo with a lot of sounds and different rooms and neat graphics with a lot of models and particles to prove me wrong on the processing part. I would really love to see such demo, no sarcasm.

[WildCat_ZA]

But think about all the processing that takes. It just is not practical for a real-time simulation, especially with this many sounds playing at once.

Getting a sound card that supports it is the best solution for that. Or you can write a HRTF algorithm and implement it in a demo with a lot of sounds and different rooms and neat graphics with a lot of models and particles to prove me wrong on the processing part. I would really love to see such demo, no sarcasm.

How do you know that "all the processing" is prohibitively expensive? Perhaps it's actually quite light-weight. We'd need an expert to know for sure. And in any case, with modern quad-core CPUs, perhaps a core could be dedicated to sound processing.

-----------------------------------------------------------------------------

As for binaural output, it has always been part of 3D sound engines. If a sound is coming from the right, it will play loud in the right ear, and will be faint in the left ear, so you know where it is coming from.

"Binaural" is not the same as "stereo". From Wikipedia (http://en.wikipedia.org/wiki/Binaural_recording):

The term "binaural" has frequently been confused as a synonym for the word "stereo", and this is partially due to a large amount of misuse in the mid-1950s by the recording industry, as a marketing buzzword. Conventional stereo recordings do not factor in natural ear spacing or "head-shadow" of the head and ears, since these things happen naturally as a person listens...

To realistically simulate a sound coming from the right, you can't just play the sound loudly in the right channel and softly in the left channel. The sound coming from the left channel needs to also be slightly muffled because, technically, it is coming from the right and should have to go around your head.

[Geister]

How do you know that "all the processing" is prohibitively expensive? Perhaps it's actually quite light-weight. We'd need an expert to know for sure. And in any case, with modern quad-core CPUs, perhaps a core could be dedicated to sound processing.
I'm a computer programmer and have already implemented something similar, and I know HRTF is much more complex calculations. Sure, it would be nice to use a core for that, but we don't all have quad-core CPUs, or dual-core CPUs. To program this to work on mono-core as well as multi-core CPUs would pretty much require triple the work. You'd need a version optimized for mono-core, one optimized for dual-core, and one optimized for other multi-core. The reason why dual-core is different is that even if they are powerful, they usually are shared and support an even load, so it's a different kind of optimization if you want to use the power of both cores. As for quad-core and up, you can, as you said, dedicate one core for that, but multi-core programming is not the simplest thing, you can ask any PS3 programmer about that. So if you want this to work with all kinds of computers (mono/dual/multi core, not talking about 32/64 bit processor optimizations), you have a lot of extra work, for just one tiny feature that only few people will be able to experience, and even fewer will even notice, as this effect does not work with everybody.

To realistically simulate a sound coming from the right, you can't just play the sound loudly in the right channel and softly in the left channel. The sound coming from the left channel needs to also be slightly muffled because, technically, it is coming from the right and should have to go around your head.
True, and it usually is, just like it would have to sound different from bouncing off a wall or an obstacle. This is neatly achieved on a hardware level by EAX, which is sadly not supported by Valve. I say that instead of suggesting that this be implemented in the games themselves, that it should be suggested Valve makes use of EAX instead of Miles, but I doubt this will ever happen.

[WildCat_ZA]

True, and it usually is, just like it would have to sound different from bouncing off a wall or an obstacle. This is neatly achieved on a hardware level by EAX, which is sadly not supported by Valve. I say that instead of suggesting that this be implemented in the games themselves, that it should be suggested Valve makes use of EAX instead of Miles, but I doubt this will ever happen.

According to the EAX article at Wikipedia ([url=http://en.wikipedia.org/wiki/Environmental_audio_extensions#Technology), EAX doesn't deal with 3D sound positioning. That's done by something like DirectSound3D or OpenAL. EAX is just for applying fancy environment effects to sounds so that the player feels like he's in a cave or a tunnel or a small room or a ventilation shaft ... etc.

Valve already seem to be doing all those cool sound effects in software inside their engine. More info here (http://source.valvesoftware.com/audio.php). They just need to add "Headphone HRTF" in there too. :) Maybe it would be a bit like HDR - something that not everyone would switch on for a year or two due to performance issues. At least people with the hardware to handle it would have the option to turn it on in the Audio menu.

[Geister]

You know what? Just got an idea. Since the engine already has DSP filtering, a quick solution to this would simply be to add a filter between the position of the two "ears". No need for complex HRTF function, just using a simple DSP effect should do it.

[WildCat_ZA]

While reading the Wikipedia article about EAX (http://en.wikipedia.org/wiki/Environmental_audio_extensions), I noticed a section about EAX emulation (http://en.wikipedia.org/wiki/Environmental_audio_extensions#EAX_emulation). Apparently, there are EAX emulation drivers for computers with only onboard audio. It's called "X-Fi MB" and I Googled the term and found this page (http://www.creative.com/oem/products/software/x-fimb.asp) on Creative's site. Unsurprisingly, it doesn't appear to be free software and it looks like it's only available for redistribution with certain pre-built PCs.

But! ... What I did notice is that the CMSS-3D feature is present in these emulation drivers, which suggests that it's not that expensive in terms of computational requirements. If they can do CMSS-3D in software with these drivers, surely something similar could be implemented into the Source engine. It's also worth noting that CMSS-3D is available on the X-Fi XtremeAudio, which doesn't even have the hardware acceleration chip on it.

My guess is that CMSS-3D is probably just an array of DSP filters that get applied to sounds based on their position in the game world relative to the listener.

[Irongrip]

Think of the posibilities:
A rocket passes over your head and you hear the hum go over you and the violent explosion behind your back. :D

Can we get a moderator to forward this to valve's sound team or something?

[WiZeNgAmOtX]

From my review of Half Life 2:

"The sound is phenomenal and when paired with a Creative X-Fi chipset, makes sound all the more exciting. Though Source games do not utilize EAX, as the engine does its own effects and special calculations, those with headphones are left something more. If one either does not use headphones or does not own an X-Fi sound card, it is recommended to ignore the following, as each sound chipset varies. Creative X-Fi uses CMSS-3D to down-mixes a multichannel signal and virtualizes it for headphones. For best results, uncheck "sync with control panel", and set your speaker setup to "headphones", as denoted in the Creative Volume Panel. Enable "MacroFX" and "ElevationFilter" under CMSS-3D, setting both to "Auto". Turn off the Equalizer. Set the Crystallizer to taste. Disable EAX effects. In-game, select under Audio, the "5.1 speakers" option; one will notice that with 5.1 enabled, the effects and vocals will be less distorted and muddy. Hollywood quality voice acting and distant orchestral scores that play out of the occasional radio tend to leave lasting impressions and the gamer wanting more."

That stated, I would definitely support binaural recording with Valve soft ware. I, rather, insist upon it.

[Stryk]

While reading the Wikipedia article about EAX (http://en.wikipedia.org/wiki/Environmental_audio_extensions), I noticed a section about EAX emulation (http://en.wikipedia.org/wiki/Environmental_audio_extensions#EAX_emulation). Apparently, there are EAX emulation drivers for computers with only onboard audio. It's called "X-Fi MB" and I Googled the term and found this page (http://www.creative.com/oem/products/software/x-fimb.asp) on Creative's site. Unsurprisingly, it doesn't appear to be free software and it looks like it's only available for redistribution with certain pre-built PCs.

But! ... What I did notice is that the CMSS-3D feature is present in these emulation drivers, which suggests that it's not that expensive in terms of computational requirements. If they can do CMSS-3D in software with these drivers, surely something similar could be implemented into the Source engine. It's also worth noting that CMSS-3D is available on the X-Fi XtremeAudio, which doesn't even have the hardware acceleration chip on it.

My guess is that CMSS-3D is probably just an array of DSP filters that get applied to sounds based on their position in the game world relative to the listener.I've worked with software that does manipulation of binaural recordings, ironically to undo and then redo the HRTF. This software isn't cutting edge technology wise by any means. In the 90's their processing was on a card. Around 2000 they moved to pure software processing. A machine from 2000 can do this in software realtime and have a bunch of realtime change on the fly additional filtering being active at the sametime. I also get the feeling there's still plenty of processing headroom left as well. A modern soundcard would be able to handle all the processing easily. A modern CPU might notice, but it would be far from all consuming by anymeans.

[Geister]

I've worked with software that does manipulation of binaural recordings, ironically to undo and then redo the HRTF. This software isn't cutting edge technology wise by any means. In the 90's their processing was on a card. Around 2000 they moved to pure software processing. A machine from 2000 can do this in software realtime and have a bunch of realtime change on the fly additional filtering being active at the sametime. I also get the feeling there's still plenty of processing headroom left as well. A modern soundcard would be able to handle all the processing easily. A modern CPU might notice, but it would be far from all consuming by anymeans.
Was is true HRTF or emulated? What I mean by this is did you use filters to emulate it (which is the common practice) or did you use the algorithm to calculate it?

Using filters you can emulate it so only sound engineers with super high quality headphones would notice the difference, for a much lower cost.

The reason why I say it is consuming resources is that it can add delays in the sound buffer (when using true HRTF, not emulation) up to 50ms, which is a noticeable difference. However, using DSP filters, this can be achieved with little to no delay.

CMSS-3D is probably, as you said, an array of DSP filters, and not the true calculation. And anyway, who wants the true calculations when you can get nearly the same effect with just a bunch of filters? (only freaks... don't look at me! lol)

I guess then a few more filters would not harm anything and make it sound better.

PS: When I say the true calculation, I mean the actual bending of the sound waves and resonance of the head which also modifies the waves sent to the ears. I guess this is only used for demos where we sometimes record our sounds and when we're happy with the location and sequence, we apply a bunch of effects and just play the recorded post-effect sounds. So there's no real-time processing, every bit of power is left to the CPU for the raytracer. =P

[WiZeNgAmOtX]

Was is true HRTF or emulated? What I mean by this is did you use filters to emulate it (which is the common practice) or did you use the algorithm to calculate it?

Using filters you can emulate it so only sound engineers with super high quality headphones would notice the difference, for a much lower cost.

The reason why I say it is consuming resources is that it can add delays in the sound buffer (when using true HRTF, not emulation) up to 50ms, which is a noticeable difference. However, using DSP filters, this can be achieved with little to no delay.

CMSS-3D is probably, as you said, an array of DSP filters, and not the true calculation. And anyway, who wants the true calculations when you can get nearly the same effect with just a bunch of filters? (only freaks... don't look at me! lol)

I guess then a few more filters would not harm anything and make it sound better.

PS: When I say the true calculation, I mean the actual bending of the sound waves and resonance of the head which also modifies the waves sent to the ears. I guess this is only used for demos where we sometimes record our sounds and when we're happy with the location and sequence, we apply a bunch of effects and just play the recorded post-effect sounds. So there's no real-time processing, every bit of power is left to the CPU for the raytracer. =P Read what I posted before in this thread. Our brain is what senses direction, not software or speaker counts. We only have two ears. Read more often. I am going to leave my comments with that.

[Irongrip]

Read what I posted before in this thread. Our brain is what senses direction, not software or speaker counts. We only have two ears. Read more often. I am going to leave my comments with that.

Did you completely misread what Geister said.
Of course the human ear will not make the difference between the filters and the real hands on approach to HRTF.

Though if Valve can implement this they could also add support for dsp filters embedded in the map files and thus the community can make extra filters for new situations.
Even if they don't make the binaural audio possible if they only add the possibility of custom dsp filters i'm sure the mod community will make them for us.

[Geister]

Did you completely misread what Geister said.
Of course the human ear will not make the difference between the filters and the real hands on approach to HRTF.

Though if Valve can implement this they could also add support for dsp filters embedded in the map files and thus the community can make extra filters for new situations.
Even if they don't make the binaural audio possible if they only add the possibility of custom dsp filters i'm sure the mod community will make them for us.
I was thinking just that yesterday when I was eating. Could be nice to hear all sounds like an alien would, or like inside the helmet of a Combine.

Also, you not only hear with your ears, but also with the resonance from your head and body by transmitting the vibrations to the inner ear, which is not possible to get with headphones as opposed to a surround sound system where the sound actually hits your body rather than only your ears. HRTF is the emulation of this, which can be quite different from one another, hence why I say the effect might not be perceived by everyone. However, if at least decent sound elevation can be achieved, it would be a good head start, but I believe that even with the best audio technology, if you have crappy headphones it just won't do it.

[WiZeNgAmOtX]

I was thinking just that yesterday when I was eating. Could be nice to hear all sounds like an alien would, or like inside the helmet of a Combine.

Also, you not only hear with your ears, but also with the resonance from your head and body by transmitting the vibrations to the inner ear, which is not possible to get with headphones as opposed to a surround sound system where the sound actually hits your body rather than only your ears. HRTF is the emulation of this, which can be quite different from one another, hence why I say the effect might not be perceived by everyone. However, if at least decent sound elevation can be achieved, it would be a good head start, but I believe that even with the best audio technology, if you have crappy headphones it just won't do it.
My head phones are Audio Technica ATH-A900 Art Phones.

[WiZeNgAmOtX]

Did you completely misread what Geister said.
Of course the human ear will not make the difference between the filters and the real hands on approach to HRTF.

Though if Valve can implement this they could also add support for dsp filters embedded in the map files and thus the community can make extra filters for new situations.
Even if they don't make the binaural audio possible if they only add the possibility of custom dsp filters i'm sure the mod community will make them for us. Soft ware emulation is not as good as hard ware emulation. Again, as I seem to need to repeat my self, the CMSS-3D is what you are speaking of. I explained it.

[WildCat_ZA]

Read what I posted before in this thread. Our brain is what senses direction, not software or speaker counts. We only have two ears. Read more often. I am going to leave my comments with that.
Soft ware emulation is not as good as hard ware emulation. Again, as I seem to need to repeat my self, the CMSS-3D is what you are speaking of. I explained it.
These posts make no sense to me, given the context of the current discussion. Both of them seem to be replies to posts made by other users, but neither of them appear to actually reply to said posts.

Video related (http://www.youtube.com/watch?v=lfJAHASV8k8)

[Irongrip]

I just noticed both VLC AND the codec thing that you configure when you install CCCP (most likely FFaudio) have the virtual headphones surround audio...

So then what will happen if you somehow made HL2 use FFaudio output and set the audio to 5.1/surround from inside of HL2?