Room Acoustics Prediction and Auralisation: Verification and Testing of New Methods in Room Acoustics Modelling.

点击翻译按钮获取中文摘要

Room acoustics simulation - modelling acoustic wave propagation within an enclosed space - is one of the fundamental applications of audio signal processing. Given pre-defined sound-source/listener locations and a surrounding geometry the resulting room impulse response (RIR) of the system can be found. Convolution with this RIR allows any audio signal to be placed in the room at the source location and auditioned by a listener placed at the receiver location. Essentially this allows any sound source to be heard within any room.There are three main applications for this technology:(1) Reverberation simulation in music production: all music is composed to be heard in a reverberant environment, whether a concert hall or an algorithm designed to simulate an optimal reverberant field.(2) Architectural Acoustics: where a building is simulated and auditioned before construction/renovation to determine the resulting acoustic quality and identify any changes that might be appropriate.(3) Virtual environment modelling: e.g. computer games, virtual reality applications and film/television postproduction, where various sound-sources are placed in and around a virtual environment with the potential for allowing interaction with the space.Although research at both York and Aalto encompasses standard room acoustics simulation methods based on geometric acoustic techniques (where sound is assumed to behave as a ray of light with the associated limitations involved with making such an assumption) most of our interests and efforts are focused on approaches that solve the acoustic wave equation directly, that therefore offer the potential for a full, complete and accurate simulation of the soundfield within an enclosed space. Recent research has included optimal grid-sampling schemes, frequency dependent diffusing boundaries, spatial encoding for receivers and source excitation strategies. Directional source encoding and real-time auralisation have also been explored, taking the best aspects of wave based and geometric approaches, and offering significant and real potential for both further research and commercial exploitation.Hence, solutions now exist for the main constituent features of any simulation - source excitation and directivity, wave propagation, boundary interaction, and receiver encoding. This is coupled with new modelling methods and the possibility of using Graphical Processing Units to speed up calculation times. Much of this work to date has been validated using simple, easy to measure objective metrics such as room mode analysis, reverberation time, polar directivity plots, boundary characteristics, etc. It therefore now seems appropriate to tackle more demanding simulations with a view to validating this approach for a broader range of problems. A number of options will be explored including simple, analytically trivial shoebox-shaped rooms, previously published data that formed the basis for a round robin study on room acoustics measurement and simulation, as well as the study of new, complex spaces based on their direct measurement, both physical and acoustic.The aim of this visit is to facilitate a benchmark study in the use and testing of new room acoustic prediction and auralisation methods for a number of specific ideal and real-world scenarios. This therefore leads to the following objectives: (1) Consolidation of code into an appropriate framework for carrying out this benchmark study; (2) design, carry out and write up testing of new room acoustic prediction and auralisation methods; (3) organise and present results at the first York-Aalto Auralisation Workshop.The project will result in a consolidation of our research codebase to more easily facilitate both this and future studies, at least one major journal publication and a workshop between York and Aalto to allow initial dissemination and feedback on our results, as well as further encourage the ongoing collaboration between our two groups.