AES Milan 2018
Poster Session P13
P13 - Posters: Modeling
Thursday, May 24, 16:00 — 17:30 (Arena 2)
P13-1 Nonlinear Real-Time Emulation of a Tube Amplifier with a Long Short Time Memory Neural-Network—Thomas Schmitz, University of Liege - Liege, Belgium; Jean-Jacques Embrechts, University of Liege - Liege, Belgium
Numerous audio systems for musicians are expensive and bulky. Therefore, it could be advantageous to model them and to replace them by computer emulation. Their nonlinear behavior requires the use of complex models. We propose to take advantage of the progress made in the field of machine learning to build a new model for such nonlinear audio devices (such as the tube amplifier). This paper specially focuses on the real-time constraints of the model. Modifying the structure of the Long Short Term Memory neural-network has led to a model 10 times faster while keeping a very good accuracy. Indeed, the root mean square error between the signal coming from the tube amplifier and the output of the neural network is around 2%.
Convention Paper 9966 (Purchase now)
P13-2 Audio Control Room Optimization Employing BEM (Boundary Element Method)—Robert Hersberger, Walters Storyk Design Group - Basel, Switzerland; Fachschule für Akustik; Gabriel Hauser, Walters Storyk Design Group - Basel, Switzerland; Dirk Noy, WSDG - Basel, Switzerland; John Storyk, Walters-Storyk Design Group - Highland, NY, USA
The Boundary Element Method (BEM) is a state-of-the art tool in many engineering and science disciplines. In acoustics, the usage of BEM is increasing, especially for low frequency analysis, since the computational effort for small to medium geometries and long wavelengths is comparatively small. While BEM is well known to give reliable results for correctly programmed room shapes, the paper at hand demonstrates that the BEM model can also respond accurately to inserted absorptive materials, and hence the method is useful for virtually prototyping the efficiency of proposed acoustical modifications ahead of actual construction.
Convention Paper 9967 (Purchase now)
P13-3 A Machine Learning Approach to Detecting Sound-Source Elevation in Adverse Environments—Hugh O'Dwyer, Trinity College - Dublin, Ireland; Enda Bates, Trinity College Dublin - Dublin, Ireland; Francis M. Boland, Trinity College Dublin - Dublin, Ireland
Recent studies have shown that Deep neural Networks (DNNs) are capable of detecting sound source azimuth direction in adverse environments to a high level of accuracy. This paper expands on these findings by presenting research that explores the use of DNNs in determining sound source elevation. A simple machine-hearing system is presented that is capable of predicting source elevation to a relatively high degree of accuracy in both anechoic and reverberant environments. Speech signals spatialized across the front hemifield of the head are used to train a feedforward neural network. The effectiveness of Gammatone Filter Energies (GFEs) and the Cross-Correlation Function (CCF) in estimating elevation is investigated as well as binaural cues such as Interaural Time Difference (ITD) and Interaural Level Difference (ILD). Using a combination of these cues, it was found that elevation to within 10 degrees could be predicted with an accuracy upward of 80%.
Convention Paper 9968 (Purchase now)
P13-4 Design of an Acoustic System with Variable Parameters—Karolina Prawda, AGH University of Science and Technology - Kraków, Poland
The number of multifunctional halls with need for acoustic adaptation aimed at many different demands of the spaces is constantly growing. The present paper shows the design of an acoustic system with variable characteristics and adjustable sound absorption coefficient that may be used in such spaces.
Convention Paper 9969 (Purchase now)
P13-5 High Frequency Modelling of a Car Audio System—Aleksandra Pyzik, Volvo Car Corporation - Torslanda, Sweden; Andrzej Pietrzyk, Volvo Car Corporation - Torslanda, Sweden
Geometrical Acoustics is a widely used room acoustic modelling method. Since GA neglects the wave phenomena and is strictly applicable for short wavelengths relative to model and surface sizes, the application for the automotive industry is still subject of research. The paper studies the feasibility of using GA for high frequency simulations of a car sound system. The GA models of a vehicle at three production stages were created based on FE models. An impedance gun was used for in-situ measurements of the properties of the car interior materials. The directivity of the midrange and tweeter speakers was measured in anechoic conditions. In subsequent simulations, various GA software settings were tested. Simulation results were verified with measurements in the car.
Convention Paper 9970 (Purchase now)