Authors:Thorborg, Knud; Futtrup, Claus
Affiliation:Scan-Speak A/S, Videbaek, Denmark
Previous circuit models of dynamic moving-coil transducers that incorporated “semi-inductors” need to be extended in order to cover those cases that failed to adequately predict performance. The improved model takes into account semi-inductance, the effects of AC shorting devices, and the impact of frequency-dependent loss in the suspensions. As shown by the good agreement between the models and measurements, only a few additional components need be added to the model. This represents a more detailed physical understanding of loudspeaker behavior.
Download: PDF (HIGH Res) (5.8MB)
Download: PDF (LOW Res) (1.6MB)
Authors:Tsingos, Nicolas; Jiang, Wenyu; Williams, Ian
Affiliation:Sound Technology Research, Dolby Laboratories, San Francisco, CA, USA; nVidia, Santa Clara, CA, USA
The rapid evolution of dedicated graphics hardware during the last decade allows for a dramatic increase in processing power for many audio applications that are unrelated to visual rendering. Modern graphical processing units can be used for audio processing in virtual reality and computer games. Their massively-parallel architecture offers 5- to 100-fold performance improvement compared to CPU implementations, as shown by such applications as audio rendering or speech recognition. Improved hardware features in the form of memory caches, multithreaded calls, better connection to the CPU, and standardized programming tools make them an attractive choice for compute-intensive audio applications.
Download: PDF (HIGH Res) (41.0MB)
Download: PDF (LOW Res) (1.3MB)
Authors:Ebem, Deborah U.; Beerends, John G.; Van Vugt, Jereon; Schmidmer, Christian; Kooij, Robert E.; Uguru, Joy O.
Affiliation:Department of Computer Science, University of Nigeria, Nsukka, Enugu-State, Nigeria; TNO, Delft, The Netherlands; OPTICOM GmbH, Erlangen, Germany; Delft University of Technology, Faculty of Electrical Engineering, Mathematics and Computer Science, Delft, The Netherlands; Department of Linguistics, Igbo and other Nigerian Languages, University of Nigeria, Nsukka, Nigeria
Models that represent objective speech quality should include the influence of the cultural background of the listener as well as the particular language being spoken. Tests that evaluated American English and Igbo, an African tone language, revealed significant differences. Igbo listeners were more disturbed by additive noise and low listening levels. The low-level parts of the Igbo tone language appear to contain more critically important information than American English. Quality speech models must therefore be tuned to match the culture and language.
Download: PDF (HIGH Res) (1.6MB)
Download: PDF (LOW Res) (443KB)
Authors:Cengarle, Giulio; Mateos, Toni; Bonsi, Davide
Affiliation:Fundació Barcelona Media, Barcelona, Spain; Fondazione Scuola di San Giorgio, Musical and Architectural Acoustics Laboratory, Venezia, Italy
The second-order components of a spherical harmonics expansion of the sound pressure field can be obtained by means of velocity transducers based on hot-wire anemometry. Such velocity transducers are small compared to acoustic wavelengths and can be mounted on a tiny rigid bar only a few millimeters wide, thereby reducing the effects of wave diffraction and acoustic shadows. Numeric simulations and experimental tests using two Microflown pressure-velocity transducers validate the proposed approach.
Download: PDF (HIGH Res) (4.4MB)
Download: PDF (LOW Res) (638KB)
Binaural audio employs “head-related” signals to create the impression of spatial realism, but whose head and ears should be represented and does it make any difference? We summarize recent research in the field, presented at AES events.
Download: PDF (10.5MB)
Author: John Krivit