Wednesday, April 27, 2011

3D Sound From Ordinary Speakers

Edgar Choueiri, a professor of mechanical and aerospace engineering at Princeton University, has developed a way to play true three-dimensional sound recordings over regular loudspeakers, such as those found in televisions and computer laptops.

In one demonstration, two microphones were placed on a dummy’s head who is made to sit in a conference table with several other people. The recorded sound from the two microphones were played back through two ordinary speakers - after processing them using Choueri's 3D filter - in another room where the listener is. The listener was able to correctly locate the position of each speaker.

Segments of the video above incorporate Choueri's 3D filter to demonstrate this phenomenon. The filter is designed to work with loudspeakers - not headphones - and can be experienced through standard computer speakers. Make sure the right and left speakers are on the correct sides

The technique may one day be used to allow 3D televisions to produce lifelike sound and to help people with certain types of hearing impairments locate noises. [via AudioLemon]

One Reddit user chips in with additional information.

This stuff has been available for about 15 years. I used to work for Sensaura. We licensed to audio chip manufacturers. All of the problems mentioned in here (cross-talk cancellation, different shapes of pinnae etc for example) have been solved a long time ago. Our technology was in pretty much every PC soundcard that was not made by Creative. It was also in the original Xbox and nForce motherboards. Our later implementation was completely CPU-based; it was all done in the driver and would work on any AC97 chipset. We also had a software implementation for consoles called GameCODA.

Then we got bought by Creative and those patents have pretty much sat in their cupboard, along with the patents from Aureal. Microsoft is also to blame for the demise of 3D audio as when Vista came out, they changed the audio subsystem so that you couldn't access audio hardware (or a custom driver like ours) through DirectX.

So if you build yourself a time-machine, go back 10 years and buy a Turtle Beach Santa Cruz soundcard, all of your 3D audio dreams will come true. In my mind, it was the best soundcard ever made and included Virtual Ear in the driver, which let you tune the 3D audio to your own ear shape. We also had MacroFX, which simulated sounds very close to the head, ZoomFX, which simulated sounds of any size instead of just a point source, a version that would add full 3D (including vertical positioning) to a surround sound setup, and a whole bunch of other marvelous stuff.

Do checkout his his comment for links to whitepapers and more information on the subject.


  1. ridiculous. this was incredibly illuminating. thank you.

    I'm a bit shocked why this hasn't become a standard for processing in hi-fi surround (or even stereo) music - let alone movie editing! maybe it's not compatible with surround speakers though - but I'm sure some configuration of it is.

    it's only time delaying and amplitude tweaking- so it's high quality processing that can be done in 24-bit editing environment, no problem.


    1. "it's only time delaying and amplitude tweaking- so it's high quality processing that can be done in 24-bit editing environment, no problem." uh, nooooooooooooo. It's projecting left channel audio + right channel cancellation (via inverted right channel) from the left speaker and right channel audio + left channel cancellation (via inverted left channel) from the right speaker. At your left ear for example, the algebraic summing of the right channel's direct (and unintended) wave and the right channel's cancellation wave effectively null such that the left ear hears only the channel intended for that ear. The effect is extremely similar to listening with fully enclosed earcup headphones, without the bother of wearing headphones. The overwhelmingly impressive playback effects demonstrated in the lab's anechoic chamber, however, are based on binaural recordings from a dummy head with mics in each ear, NOT typical "stereo" recordings which have already muddled point source sounds so much.


Popular Posts