Untitled Document

Projects in AMBISONICS

EE/Mu 107 Winter Term:
Our own Ambisonic Setup

EE/Mu 107 Spring Term:
B+format
Resources/Links

After learning about the theory and advantages over standard surround sound, I wanted to listen to Ambisonics for myself. I discovered some free computer software that could play B-Format recorded material. This lead to building a computer and setting up the MusicLab for our own demonstration.

what is Ambisonics?
Ambisonics is a surround sound system of recording and playback that uses the psycho-acoustic model of human directional hearing to provide accurate, convincing localizations of sounds in a three-dimensional space. The Ambisonics system is superior to the mainstream offerings of 5.1 Surround Sound, Dolby Digital, etc. on several levels. First, it provides dense 3D (or periphonic) images, whereas others are 2D and have "holes". Second, the data stream is encoded into UHJ format, which degrades nicely, meaning it provides full periphonic playback when one has a decoder, but still plays properly on stereo and mono systems without special hardware, something that "the others" cannot do. Further, only four channels of information need to be transmitted for full periphonic playback, whereas five or more are needed for the other systems, which have at best, a 2D image. More details here.

how was it set up?
The setup involved the construction of a computer with eight channels of soundcard output, the acquisition of four stereo amplifiers and eight speakers, the construction of four speaker stands, and cable fabrication.

The computer is a basic Pentium III 800MHz system running Windows 98. The software I used is a program written by Angelo Farina called AmbisonicPlayer. It allows the use of multi-channel soundcards or multiple soundcards to drive eight channels in a cube format (one speaker per vertex). I used four Crystal Audio stereo soundcards that I purchased for $10 each.

Four NAD 3020 stereo amplifiers were used for playback. Two were already available in the MusicLab, and two others were kindly lent for use from Peter Sutheim.

Eight NHT 1.1 two-way speakers made their way into the setup, two from the MusicLab, four borrowed from James Boyk, and two from Leon Bellen.

Four speaker stands were constructed. They consist of 1/2" iron pipe lengths and connectors, 3/4" MDF shelving, and a 3/4" x 24" diameter round plywood base. A guy wire was connected from the base to the top of the stand for rigidity, and the shelves were sprayed with semi-flat black paint. Each stand is approximately 80" tall, with an effective 72" separation in height between the upper and lower speakers. Cost per stand was $45.

Cables from the computer to the amp were fabricated using high quality MediaTwist twisted pair with stereo mini jacks on one end and RCA plugs on the other. Two cables of eight conductors each were used. $15 went for ends, and all was soldered by hand. Speaker wire consisted of quality CL3-rated 14 gage/4 conductor cable, one per stand. 145' was used, which cost $70

what were the results?
The initial effect was disappointing, but this was due to the improper location of speakers in the room (a non-cubic spacing dictated by room dimensions was used). The software was configured for a strict cube setup. Once the speakers were relocated, the results were quite convincing. I was impressed by the sense of really "being there."

It was discovered that the software was not configurable for custom speaker placement, a serious limitation. Further, it allowed only rudimentary playback capabilities, falling far short of the basic offerings from analog processors, which allow image manipulation via rotation and tilt.

The excellent quality of the speakers and amps revealed a significant amount of noise coming from the computer. I guess this can be expected from commodity soundcards.

what's next?
I am interested in continuing research in the area of Ambisonics, and plan on writing my own software playback program that addresses issues not met by the basic AmbisonicPlayer. It will allow arbitrary speaker location, pan and tilt capabilities, and the ability to decode standard UHJ recordings, which are available already on many CDs and LPs.

Better soundcards for higher quality output signals are needed, as well as a permanent set of speakers and amps.

advice…
Do a lot of reading. A wealth of knowledge is available on the www.ambisonic.net website. Look for an Ambisonics setup available for demo. The listening experience is definitely worth the effort.

questions?
email joshua@caltech.edu for information regarding this project.

B+format
Dr. Thomas Chen has developed an extension of the B-format, which he calls "B+format" for an optional enhancement to the Ambisonics listening experience. In a nutshell, Soundfield microphone recordings tend to be better suited for capturing environmental sounds (like room acoustics) than direct source material. In the B+format, an additional standard stereo recording is made simultaneously with a standard Soundfield mike, and is usually placed closer to the target source being listened to. Instead of the standard four channels in a first order Ambisonic recording, the stereo pair is added for a total of six. At the decoding stage a weighted blending of the recordings may be made available to the listener to accentuate the main source within the environment.

Click here for my experiences at Dr. Chen's listening setup in Stockton, CA.

Links