2004 July/August, Volume 52 Number 7/8

CONTENT

PAPERS

Predicting the Perceived Quality of Nonlinearly Distorted Music and Speech Signals
Chin-Tuan Tan, Brian C. J. Moore, Nick Zacharov, and Ville-Veikko Mattila   699
An analytical model predicts the perceived quality of speech and music when subjected to various forms of nonlinear distortion, thus providing the ability to evaluate audio transducers and transmission systems. The model incorporates the filtering produced by the auditory system so that the perceptual consequences of spurious spectral components are correctly evaluated. By cross-correlating each spectral bin of the input and output signals, the measure of distortion, called Rnonlin, accurately predicts the subjective ratings of listeners. This approach improves on the earlier metric, called Distortion Score, which was less accurate for real transducers.

Multi-Actuator Panels (MAPs) as Loudspeaker Arrays for Wave Field Synthesis
Marinus M. Boone   712
Distributed mode loudspeakers become suitable for wave field synthesis when multiple exciters are embedded in a single panel. In contrast, a large number of loudspeaker panels is not efficient and has low-frequency limitations. As long as the initial part of the impulse response of individual drivers is phase consistent, deviations from an ideal response can be corrected with filtering. Increasing the internal damping by applying a foam board panel was sufficient to achieve accurate wave front curvature as would be obtained from real sources. Multi-actuator panel arrays of loudspeakers are lightweight and thin and can be readily mounted on wall surfaces.

Hierarchical Automatic Audio Signal Classification
Juan José Burred and Alexander Lerch   724
The advantages and disadvantages of using a hierarchical system for analyzing and classifying audio file content are explored. Such systems are analogous to content analysis for text material. In the hierarchical approach, top-level tree decisions, such as distinguishing between speech and music, approach 95% accuracy, while the more subtle distinctions lower in the tree, such as genres of chamber music, remain at a modest 50%. The authors provide an extensive review of the state of the art in order to place their research into perspective.

ENGINEERING REPORTS

A Frequency-Domain Approach to Multichannel Upmix
Carlos Avendano and Jean-Marc Jot   740
Because of the wide availability of mulitchannel reproduction equipment and the limited availability of music recorded for formats other than stereo, upmixing a two-channel recording into multichannel format (for example 5.1) would be very useful. By comparing the two stereo channels, it is possible to identify and unmix amplitude-panned sources and to identify and extract background spatial ambience. This allows the redistribution of spatial information and the creation of additional channels to be rendered by the multichannel equipment. The algorithm is based on analyzing the two channels in the time-frequency domain and then observing the degree of similarity and correlation between them.

STANDARDS AND INFORMATION DOCUMENTS

AES Standards Committee News   750
Digital audio connectors; synchronization; transfer technologies; listening tests; shielding; file interchange; Internet audio

FEATURES

116th Convention Report, Berlin   756
    Exhibitors   772
    Program   776
117th Convention Preview, San Francisco   828
    Exhibitors   831
    Exhibit Previews   835
Education News   860

DEPARTMENTS

News of the Sections   862
Sound Track   866
Available Literature   867
Upcoming Meetings   868
Membership Information   868
Advertiser Internet Directory   871
Sections Contacts Directory   873
AES Conventions and Conferences   880

FPO For the cover:

2004 July/August, Volume 52 Number 7/8

spine: 2004 July/August, Volume 52 Number 7/8