Use of Logistic Regression Models as a Supervised Learning Algorithm to Identify Impulsive Sounds in Monitored Sound Data

Authors

  • Harry Ao Cai HGC Engineering, CA

Abstract

Impulsive sounds, characterized by their transient nature, often pose challenges in sound monitoring applications such as environmental noise assessments, where impulsive sounds need to be identified and processed separately from other impulsive and non-impulsive sounds. Several methods exist to identify impulsive sounds, such as through listening to recorded audio or a manual examination of the logged frequency-spectral data. This paper presents of the use of logistic regression models, implemented as a supervised learning algorithm, to identify impulsive sounds from monitored sound data. Logged spectral sound pressure data from sound level meters with were used as input. A segment of the data was pre-labelled for the impulsive sounds of interest and was used to train a logistic regression model to identify the same impulsive sounds in other data. This method aims to automate some of the analysis procedure required for handling large volumes of spectral data involving impulsive sounds.

Additional Files

Published

2023-10-09

How to Cite

1.
Cai HA. Use of Logistic Regression Models as a Supervised Learning Algorithm to Identify Impulsive Sounds in Monitored Sound Data. Canadian Acoustics [Internet]. 2023 Oct. 9 [cited 2024 Jul. 14];51(3):70-1. Available from: https://jcaa.caa-aca.ca/index.php/jcaa/article/view/4098

Issue

Section

Proceedings of the Acoustics Week in Canada