|1996 DECEMBER, VOLUME 44 NUMBER 12|
Presidents' Message Tim Shelton and Elizabeth A. Cohen p. 1083
PAPERSCo-talker Separation Using the 'Cocktail Party Effect'
Yuchang Cao, Sridha Sridharan, and Miles Moody p. 1084
The cocktail party effect is observed using two channels. These data are processed in an artificial neural network speech classifier. After a learning period during which the classifier becomes familiar with the desired speaker's voice characteristics, the classifier then controls the parameters of a modified Wiener filter to cancel out the interference. The simulation results have shown that the system consistently provides speech of good quality and intelligibility.
An Integral Limitation upon Loudspeaker Frequency Response for a Given Enclosure Volume
Jeffrey Harrison p. 1097
The tradeoff between loudspeaker enclosure volume, efficiency, and frequency response is stated in a key integral expression, which rules out the existence of a "magic" alignment for a superior performance. This had remained a possibility when using the traditional figure-of-merit approach. Multiple-cavity loudspeaker enclosures of the same volume show no efficiency advantage over single- cavity systems in this analysis.
Compatibility Matrixing of Multichannel Bit-Rate-Reduced Audio Signals
Warner R. Th. ten Kate p. 1104
Matrixing of bit-rate reduced audio signals may cause audible artifacts in both downmixing and demixing operations. The effect is more apparent in demixing since there is a strong correlation between the compatible and auxiliary signals. It is possible to preclude the occurrence of artifacts by using the same correlation between compatible and auxiliary signals.
Sound Reinforcement by Wavefield Synthesis: Adaptation of the Synthesis Operator to the Loudspeaker Directivity Characteristics
Diemer de Vries p. 1120
A linear array of loudspeakers that have arbitrary but identical directivity characteristics can be used to generate a reinforced soundfield with natural time and space properties by using wavefield synthesis based on Rayleigh's representation theorem. Individual loudspeakers in the array that have non-identical characteristics cannot, or can only partially, be corrected for by driving them with a signal that would be appropriate for an array of such loudspeakers.
STANDARDS AND INFORMATION DOCUMENTSAES Standards Committee News p. 1132
Omission in draft revised AES11-xxxx; digital interface user format proposal; SC-02 Copenhagen meeting
FEATURES101st Convention Report p. 1134Exhibitors p. 1150AES News: Formation of Latin American Region p. 1178
Program p. 1154
1996/1997 AES International Sections Directory p. 1179
Call for Awards Nominations p. 1206
Call for Nominations for the Board of Governors p. 1207
AES 8th Tokyo Regional Convention, Call for Papers p. 1210
AES 103rd Convention, New York, Call for Papers p. 1212
Bylaws: Audio Engineering Society, Inc. p. 1214
Index to Volume 44 p. 1218
DEPARTMENTSNews of the Sections p. 1198
Sound Track p. 1202
New Products and Developments p. 1204
Available Literature p. 1205
Upcoming Meetings p. 1205
Membership Information p. 1208
In Memoriam p. 1209
Section Contacts Directory p. 1228
AES Conventions and Conferences p. 1232