Sound system lets listeners move

By Kimberly Patch, Technology Research News

Real-life audio is complicated -- our brains can keep track of sounds as they move around us even when we are moving around as well.

Researchers from the University of California at Davis have designed a relatively inexpensive spatial sound system that accounts for the user's movement as it creates the impression of sounds in the space around him.

The headphone-based system could be used in remote listening situations like teleconferencing, surveillance and teleoperation to allow people to hear events as they happen, said Richard Duda, a research engineer at the University of California at Davis. It could also be used to make spatial audio and video recordings, and for immersive interactive multimedia like computer games, augmented reality systems, and industrial and military training, he said.

People locate sounds in three dimensions -- azimuth, or left versus right, elevation, or up versus down, and range, or near versus far, said Duda.

The primary cues for azimuth are the loudness and timing differences of a sound arriving at the right versus left ear. The primary cue for elevation is how the outer ears change the sound. And the primary cue for range is loudness -- the difference in energy of the sound that arrives directly versus sound that arrives later reflected from environmental surfaces. More subtle cues include torso reflections, familiarity with a sound source, and visual cues.

Binaural recordings, which reproduce sound in space, are routinely made by placing microphones in the ears of a dummy head. This reproduces sound at the location where it is heard and accounts for the changes in sound produced by the shape of the head.

The trouble with this method, however, is that it is only accurate when the user is facing the same direction as the dummy.

People are sensitive to how sound cues change when they turn their heads, said Duda. "When the cues all change consistently, the perception of a well-defined spatial location for the source is strong; if the cues are inconsistent, the perception can be so vague that the listener has no idea where the source is," he said.

One way to address this is to use a head tracker to determine the location of the user's ears and a servo mechanism that rotates the dummy-mounted microphone array in real-time to match the user's head movements.

The researchers' system overcomes the need to move the dummy head by embedding a series of microphones into the dummy head and sampling from the nearest microphone as the listener rotates her head, said Duda. "It... occurred to us that the dummy head was really only sampling the sound field in space, and that instead of turning the dummy head physically it should be sufficient to sample of the sound field with multiple microphones around the head," he said.

In effect, this puts virtual copies of the listener's ears in the sound field and moves them as the user moves his head in order to capture the dynamic cues that are missing in conventional binaural recordings, said Duda.

Key to method was finding a way to blend, or interpolate, signals from two microphones when an ear was between microphones, said Duda. "The key technical question is how to interpolate the signals without introducing spectral distortion or requiring an unaffordable number of microphones."

The most obvious method would have required at least 128 microphones to provide CD-quality sound. The researchers found a way to use eight microphones for speech and 16 microphones for music. "Our breakthrough came from the recognition that because humans are not sensitive to differences in arrival time above about 1.5 kHz, we only had to interpolate the low-frequency components."

The researchers' recording device also has a torso. Removing the torso not only changes the perceived elevation of a sound source but also, surprisingly, the perceived distance of a source, said Duda.

The researchers' prototype doesn't possess outer ears, which for some people causes a significant shift in the apparent elevation of sounds, particularly for sounds in front, said Duda. "Or near-term goals are to understand why some people experience greater shifts than others, and to develop procedures for correcting these problems."

The researchers' ultimate goal is to create a mixture of real and computer-generated sounds, images and other stimuli to provide a compelling experience of being actually present in a remote or synthetic environment, said Duda.

Although sound is of central importance in human communication in general, including telephones, radio and television, it is woefully underused in human-computer interaction, said Duda. Especially with the advent of portable systems, sound is likely to become more important in human-computer interaction, he said.

The basic audio system is ready now, said Duda. The researchers have a rudimentary audio/video demonstration working in the lab, and could produce a low-to-moderate-quality practical audio/video application within a year, he said. High-quality audio-video applications could be practical within three years, he added.

Duda's research colleagues were V. Ralph Algazi and Dennis M. Thompson. The researchers presented the work at the 116th Audio Engineering Society Convention in Berlin, Germany May 8 to 11, 2004. The research was funded by the National Science Foundation (NSF).

Timeline:  > 1 year, 3 years
Funding:   Government
TRN Categories:  Multimedia; Signal Processing
Story Type:   News
Related Elements:  Technical paper, "Motion-Tracked Binaural Sound," presented at the Audio Engineering Society 116th Convention, May 8-11, 2004 in Berlin, Germany




Advertisements:



August 11/18, 2004

Page One

Projector lights radio tags

Cell phone melds video and data

Sound system lets listeners move

Chips measure electron spin

Briefs:
Twisted fiber filters light
Shifty tiles bring walking to VR
Speck trios make secret codes
Single gold atoms altered
Pen writes micro wires
Design eases nano locomotion

News:

Research News Roundup
Research Watch blog

Features:
View from the High Ground Q&A
How It Works

RSS Feeds:
News  | Blog  | Books 



Ad links:
Buy an ad link

Advertisements:







Ad links: Clear History

Buy an ad link

 
Home     Archive     Resources    Feeds     Offline Publications     Glossary
TRN Finder     Research Dir.    Events Dir.      Researchers     Bookshelf
   Contribute      Under Development     T-shirts etc.     Classifieds
Forum    Comments    Feedback     About TRN


© Copyright Technology Research News, LLC 2000-2006. All rights reserved.