Abstract:

Claims about the subjective performance of consumer audio products are rarely supported by psychophysical data. Advertising hype and anecdotal reports from product reviewers are typically used in place of scientific evidence. With 3-D audio products, however, research preceded product development, and existing psychophysical data address many issues related to potential product performance. These data suggest that integrating high-fidelity 3-D sound into multimedia systems is likely to be a formidable challenge. One problem is that the acoustics of the listening environment are intractable unless the listener is constrained in some way (e.g., forced to wear headphones). Another obstacle is that even if the acoustics are handled properly, large individual differences preclude generalizations about the costs/benefits of engineering compromises, such as the use of nonindividualized spatial processing. Perhaps most important, the data suggest that the perceptual errors likely to be produced in any less-than-perfect 3-D audio system are not subtle; sound images fail to externalize or are heard in the rear when they are intended for the front. These problems underscore the importance of using rigorous psychophysics to evaluate the perceptual adequacy of any multimedia sound system. [Work supported by NASA, ONR, NIH, and Rockwell Semiconductor Systems.]