More on 3D positional sound

On a Sound Blaster Live, it was basically the same story. Once set to "two-speaker mode" or "headphones" in the Speaker tab, HRTFs are automatically applied to DirectSound 3D streams. I couldn't position WAV/MIDI sounds in Windows anymore, but start the included Unreal Special Edition, and the HRTFs pop into action. I used a pair of headphones since they were universally recommended for two-driver surround. The result? Well, as I walked and flew around the now-famous "waterfall test," I did notice the sounds muffling as I turned my back to them, and as I moved up and down. Besides the changes in clarity however, I certainly wasn't convinced that I was in a real 3D environment.

On the MX300, the effect was more pronounced and admittedly more realistic, but to a very small extent. Listening to the SBLive and the MX300 side by side, I'd have to say that Aureal did a better job of implementing virtualized 3D sound. I personally don't care for the 2-speaker method myself (and found both cards somewhat lacking in this regard), but if your focus is on 2-speaker 3D sound, I would say the Monster Sound is the better choice in that regard. Just don't expect to be knocked off your feet.

What this all comes down to, is that at least for me, at that truly convincing 3D positional sound has to come from a speaker array. Four speakers are the minimum required, and that's the arrangement supported by both the SBLive and the MX300. Instead of springing for an all-around cheap set of surround speakers, I've kept my trusty CS Microworks in front and "liberated" Thresh's pair of Microworks for the rear. Using each card's included applets, I started by playing an MP3 and dragging it to various 3D positions. The SB Live performed adequately; to me it sounded as if it was linearly mapping per-speaker volume based on distance from the listener. This was fine for sounds which appeared to the far corners of the sound stage (where only one speaker would be noticeably active), but for a sound placed directly behind the listener, all four speakers would be active at nearly the same volume. This was obviously meant to convey to the listener that the sound was very close by, but in the end it pretty much sounded like it was coming from "somewhere in the middle" rather than expressly from the rear. I got this impression from the "Live Experience" demonstrations as well - it was possible, but not easy to positively identify sound sources and locations. I think a sound coming from right behind my head would sound more realistic with rear speakers at full volume and front speakers completely quiet, but it's definitely a trivial task in terms of programming. Perhaps an updated patch/driver can fix this.

Sound Placement for Surround

The MX300's drivers handled 3D sound via four speakers in a more convincing fashion. The sound applet included allowed for wave files to move around a virtual room, demonstrating real-time Doppler-shifts as the source whizzed by. It was interesting, but I couldn't say entirely realistic (I haven't had that many close calls with 60Mph boom boxes flying past my head, and there was no "ambulance" sound for a "real-world" test). It even included a vertical height slider to makes sounds appear from above or below. Needless to say, the HRTFs in an external speaker environment in addition to trying to simulate up/down positioning was less than useless, at least to me. Sure it sounded different as you moved the slider, but did the sound appear to come from above? Not at all. A3D uses HRTFs for the front speakers in order to enhance the side-positioning of sounds (which DOES work well), and uses the rear speakers for additional sound cueing. The A3D Room applet (which, by the way is the FIRST first-person 3d program I've seen to come with an inverted-T setup with STRAFELEFT/STRAFERIGHT controls by forward/backwards) convincingly rendered sounds to the front, rear, and sides, as well as demonstrated the much-prized occlusions and reflections in a 3D environment. I think Aureal hit the mark a little closer for rendering realistically placed 3D sound, either through the use of subtle HRTFs as side-positioning aides or the more discrete positioning of sounds through individual speakers. Again, they're going to need to do a lot of work for up/down, and for both cards, the framework is there and it all comes down to at this point is tweaking.