Simulação da resposta acústica ao impulso em ambientes tridimensionais utilizando computação paralela

Abstract

The acoustic characterization of environments is an important factor for project development and construction of acoustically better environments, such as concerts, classrooms, theaters, and even houses. The obtained acoustic parameters can be valued as reverberation time or even the room impulse response (RIR), which describes how the energy decays in the environment, as a function of time. These obtained parameters are related to elements like speech intelligibility in the environment. The goal of this work is to provide a program able to perform computations to calculate this impulse response, considering that the energy absorption is dependent on sound frequency, using parallel computing and a high-level programming language. For this, it was developed a program with Python language, using libraries like NumPy and SciPy for performing general mathematical operations, and the framework Apache Spark for performing the algorithm parallelization. It used Fourier transform properties for energy absorption and final response calculation. The environment geometry and parameters, as well as to object positions, are given to the algorithm through input files, making simple environment customization. The developed algorithm enables the obtainment of the impulse response of any three-dimensional environment, in parallel, consequently faster than non-parallel methods, besides demonstrating that this response is similar to the analytical base response.