Sound classification using machine learning and neural networks
Author(s):
Pooja R K, Srishti Shetty, Suhani M, Janardana D R
Keywords:
classification, feature extraction, random forest, multi class SVM
Abstract
Classification of sound automatically has been a growing field in research. Researches are mostly performed on the sonic analysis on environment sounds because of its various applications to a large scale content based multimedia indexing and retrieval. These researches are mostly focused on music or speech recognition. The Urban Sound dataset created by Justin Salamon, Christopher Jacoby and Juan Pablo Bello in 2014 is one of the few free large sound dataset. Classification of sound data is done using feature extraction. The features of sound data cannot be conveyed in vector forms such as other type of data like images and texts. Hence, the feature extraction for sound data is less unequivocal. Two categories of feature extraction techniques are applied namely, signal characteristic feature extraction and Time series feature extraction. The validness of distinct models on each method, including tests of Random Forest, Naive Bayes, Support Vector Machines and Neural Network architectures which includes deep neural network, convolutional neural network and recurrent neural network. After implementing the machine learning techniques and neural networks we are able to classify different sounds.
Article Details
Unique Paper ID: 146528
Publication Volume & Issue: Volume 4, Issue 12
Page(s): 783 - 786
Article Preview & Download
Share This Article
Join our RMS
Conference Alert
NCSEM 2024
National Conference on Sustainable Engineering and Management - 2024