FRA-RIR: Fast Random Approximation of the Image-source Method

TL;DR

FRA-RIR introduces a rapid, device-independent approximation method for generating realistic room impulse responses, enhancing speech system training efficiency and performance without specialized hardware.

Contribution

It proposes a novel random approximation approach to accelerate the image-source method for RIR simulation, enabling faster and hardware-independent data generation.

Findings

FRA-RIR significantly outperforms existing ISM-based tools in speed.

Training speech denoising models with FRA-RIR improves real-world RIR performance.

The method is easily implementable in Python and suitable for on-the-fly data generation.

Abstract

The training of modern speech processing systems often requires a large amount of simulated room impulse response (RIR) data in order to allow the systems to generalize well in real-world, reverberant environments. However, simulating realistic RIR data typically requires accurate physical modeling, and the acceleration of such simulation process typically requires certain computational platforms such as a graphics processing unit (GPU). In this paper, we propose FRA-RIR, a fast random approximation method of the widely-used image-source method (ISM), to efficiently generate realistic RIR data without specific computational devices. FRA-RIR replaces the physical simulation in the standard ISM by a series of random approximations, which significantly speeds up the simulation process and enables its application in on-the-fly data generation pipelines. Experiments show that FRA-RIR can not only be significantly faster than other existing ISM-based RIR simulation tools on standard computational platforms, but also improves the performance of speech denoising systems evaluated on real-world RIR when trained with simulated RIR. A Python implementation of FRA-RIR is available online\footnote{\url{https://github.com/yluo42/FRA-RIR}}.