Explosion sound classification using machine learning method based on audio features

Abstract

This study focuses on the classification of gunshot sounds using multiple audio features and machine learning methods. The gunshot sound samples are converted into spectrograms and processed using Support Vector Machine (SVM) for classification. The model was trained on a dataset of 851 audio files from 8 different gun types. Using a combination of audio features along with data preprocessing techniques, our SVM model achieved 95.32% accuracy in classifying different types of gunshots. The model also demonstrated good performance with real-world data, though with lower confidence levels due to environmental noise. This study provides an effective method for gunshot classification in defense security surveillance systems and sound forensics applications.