This article offers a brief explanation of the effectiveness of 3D sound and an overview of 3D sound and reverb.

The 3D sound and reverb engine your soundcard use determines what APIs (e.g. DS3D, A3D 1.0, EAX 1.0) are supported and more importantly the accuracy of the positioning of the sound sources and the quality of the reverb. CPU usage, number of 3D sound streams and the basic sound quality can differ quite significantly between soundcards/chips using the same 3D sound and reverb engine, but the positioning and reverb quality remains the same as long as they use the same version of the engine. The most common improvement in the later versions of the sound engines the last couple years have been improved reverb quality and support for EAX 1.0 and 2.0, while the improvements in the 3D sound engines have been more subtle.

In the PC market only 4 companies’ 3D sound and reverb engines have seen widespread use. The companies are (or were) QSound, Sensaura, Aureal and Creative Labs. Of those, QSound’s Q3D and Sensaura’s S-3DPA 3D sound/reverb engines are the two that were available for any chip company to license while Creative Labs 3 3D sound/reverb engines are only available as part of the drivers for their cards. The 3D sound and reverb engine that Creative Labs uses for the Live cards, what we will refer to as the Live 3D sound/reverb engine, is vastly different from the more basic 3D sound and reverb engine they offer for the ES137x based cards (e.g. SoundBlaster 16PCI, SoundBlaster PCI64, SoundBlaster PCI128). The Audigy engine is an enhancement of the Live engine but is not as radical an improvement as the step from a PCI64/128 to a Live. Aureal’s A3D engine (Editor's note: and now Sensaura's) is now the property of Creative Labs.

So on to the 3D Sound and Reverb Overview

3D Sound

The approach to create the illusion of 3D sound for headphones is in all cases based on Head Related Transfer Functions (HRTF for short). HRTF filters are developed by putting microphones in a person's ears (or rather in a lot of people's ears to create an average) and moving a sound source around him so you know exactly what each ear hears for each position of the sound source. They use that info to create the 3D sound algorithms used for headphones.

It's important to remember that even though HRTFs are used by practically all 3D audio companies for headphones, there can be a significant difference in how good they are on placing the sound in the 3D space. How well it will work for you can be quite different than what a reviewer thinks.  For that reason it's a good idea if at all possible, to try the soundcard out before you buy it, or at least be able to return it if it doesn't work like you thought it would or should. In most cases we have found elevation cues to be the best when using headphones with rear placement also being very good but placement in front of you fairly poor. In actual games the poor placement in front of you is often not that noticeable since the visual cues are then very strong.

For 2-speakers, all but QSound use HRTFs together with cross-talk cancellation. The cross-talk cancellation is required since HRTF assumes the left ear doesn’t hear what the right speaker produce, which of course not accurate when using speakers. How effective the cross-talk cancellation is depends a lot on room geometry and the positioning of the speakers but in no case is it as accurate as headphones. This is especially noticeable on elevation cues, which work very well with headphones but are fairly weak, if even noticeable, on 2-speakers. Placement in front of you and to your sides is very good with most 2-speaker algorithms while rear placement is, just like elevation cues, weaker. Unlike with elevation cues, for the most part we have found rear placement to be effective both in terms of offering a more immersive sound experience as well as offering helpful clues where your enemies are. However we don’t think any solution is even close to 4-speakers when it comes to rear placement. We have about the same impression of QSound’s different approach to 3D sound on 2-speakers.

QSound’s Q3D uses a different method than cross-talk cancellation to produce 3D sound for 2 speakers (called Q1). While the rest use HRTFs together with cross-talk cancellation (a must for it to work well) QSound decided to develop audio filters specifically for 2 speakers and they performed more than 550,000 listening tests before they finally were happy with the filters. In the QSound section we will go into more detail why QSound feels their approach to be better.

When using 4-speakers the 3D sound engines can either use an approach similar to what they do with 2-speakers or just simple volume panning. In other cases there is a combination of both. The advantage of using a similar approach as they do with 2-speakers is that you still get some elevation cues and more importantly better side placement than the engines that only use volume panning. Having said that we have found all 4-speaker solutions to offer a good 3D sound experience that excels on front and rear placement with the elevation cues being the weakest point, significantly behind what headphones can offer.

In a typical desktop configuration the 5.1 solutions don’t offer a major, if any, advantage compared to 4-speaker solutions and in some cases are even worse. In short the only case it offers a significant benefit is those scenarios where a phantom center becomes less effective.  The main cause of an ineffective phantom center is not having the proper spacing between the listener(s) and the two front channels.  By proper spacing we don’t just mean the distance between the speakers but also the distance they sit in front of the listener(s)  The most common scenario for this is with console gaming, where there are often multiple users so at least one won’t be seated in the sweet spot for a phantom center. However, even in a single listener desktop scenario there are reasons why some users may want to sacrifice some stereo imaging and instead depend on the combination a 5.1 algorithm and a real center channel speaker for anchoring channel material. You can find more details about 5.1 setup and pros and cons of 5.1 gaming in our 5.1 Gaming article (not available yet).

One other point of note that many people go astray on is games do not have to be specially written beyond supporting DS3D in order to take advantage of 5.1 sound cards 3 stereo outputs.

The new DD 5.1 encoding offered by the XBox and some of nVidia’s nForce chipsets are just a way to get the 5.1 sound to the speaker system. That is from a 3D sound perspective, it’s no different than connecting the soundcards 3 stereo outputs to your speakers 3 stereo or 6 mono inputs. There are no audio quality benefits either, in fact it can technically only hurt the audio quality. The benefit is just for those who own a 5.1 system with a Dolby Digital Decoder that lacks separate 5.1 inputs.

Reverb

The reverb engines tries to recreate the sound of a specific environment. When comparing the reverb quality of the different engines you need to take into account consistency, accuracy, how strong it is, if it’s possible to tweak the strength and the very subjective how good it sounds. Where we have noticed the most significant difference are the consistency, strength and possibility of tweaking. With consistency I refer to the fact that while it may sound very good in 7 out of 10 games relative to another card, it may sound horrible in 1 and be hardly noticeable in 2.   What makes that inconsistent is those 3 games may sound perfectly fine on another card using a different engine or different version of the same reverb engine. Unlike 3D sound we haven’t in most cases noticed any major difference between the output modes, that is it sounds about the same with 2-speakers as with 4-speakers. The reason I say in most cases is we have come across some games where it can differ significantly due to bugs.

APIs

To access the 3D sound and reverb engines the game must use an API that’s supported by the engine. All of the engines mentioned earlier support DS3D. They also all support A3D 1.0 either directly or by translating the calls to DS3D in Windows 9x/ME. Several lacks A3D 1.0 support in Windows XP/2000. The A3D 2.0/3.0 APIs are only directly supported by the A3D engine but the A3D 2/3 SDK also offer DS3D and even EAX support as an option for the game developer.

Moving on to the reverb APIs, all support EAX 1.0 in the last version of the engine. EAX 2.0 is supported by the Live and Audigy engines and the later versions of the QSound and Sensaura engines. Only the Audigy supports the new EAX AdvancedHD.  If your soundcard doesn’t support the API the game may use a software 3D sound/reverb engine or more commonly, stereo panning. You can can read our API article for more details on the 3D sound and reverb APIs as well as some of the better-known software engines.

On to Aureal's A3D Engine