Sound-based shape classification in an anechoic box using AI algorithm

• Main Authors: Nakul, Sanwlot, Akhil, V. M.
• Format: Conference Proceeding
• Language: English
• Subjects: Algorithms; Audio data; Audio signals; Classification; Feature extraction; Kernel functions; Machine learning; Microphones; Signal processing; Sound waves; Support vector machines
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0234744

This study focuses on identifying shapes by sound waves in anechoic box using feature extraction and classification techniques. The experiment setup involved speakers at both sides of the anechoic box and microphone at center. Geometric shapes made of MDF (medium density fiberboard) were placed between speakers and microphone. Feature extraction technique involving root mean square, standard deviation and variance were extracted from recorded audio signals and machine learning algorithms such as (svm) support vector machines, knn, naive bayes, ensemble, kernel and linear discriminant were used for classification of data. Result demonstrated an impressive accuracy rate of approximately 99% and accurately categorized the signals based on geometric shapes. These findings have significant implications for signal processing and classification research, providing insights into the potential applications of these techniques in real-world scenarios.