Neural network for multi-exponential sound energy decay analysis

Loading...
Thumbnail Image

Access rights

openAccess

URL

Journal Title

Journal ISSN

Volume Title

A1 Alkuperäisartikkeli tieteellisessä aikakauslehdessä

Date

2022-08-12

Major/Subject

Mcode

Degree programme

Language

en

Pages

13
942-953

Series

Journal of the Acoustical Society of America, Volume 152, issue 2

Abstract

An established model for sound energy decay functions (EDFs) is the superposition of multiple exponentials and a noise term. This work proposes a neural-network-based approach for estimating the model parameters from EDFs. The network is trained on synthetic EDFs and evaluated on two large datasets of over 20 000 EDF measurements conducted in various acoustic environments. The evaluation shows that the proposed neural network architecture robustly estimates the model parameters from large datasets of measured EDFs while being lightweight and computationally efficient. An implementation of the proposed neural network is publicly available.

Description

Keywords

sound energy decay analysis, multi-exponential sound decay, reverberation time, neural network

Other note

Citation

Götz, G, Falcon Perez, R, Schlecht, S & Pulkki, V 2022, ' Neural network for multi-exponential sound energy decay analysis ', Journal of the Acoustical Society of America, vol. 152, no. 2, pp. 942-953 . https://doi.org/10.1121/10.0013416