Document Type

Honors Thesis

Abstract

Accurate and efficient audio classification algorithms have significant applications in the modern world. Song recognition and speech recognition apps rely on complex audio processing software. Recent research in this field has focused primarily on the application of artificially intelligent systems to solve difficult audio processing problems where strictly rule-based algorithms would be untenable. The problem to be investigated in this project is a much simpler binary classification scenario, but with a much smaller sample set than is typically used for neural network training as well as limited processing resources. It will be determined whether in such a scenario a neural network approach will still outperform a strictly rule-based algorithm. Each implementation is given identical sound samples to test against. Samples include a collection of desired and undesired sounds. Accuracy is measured as a percentage of how many test samples are correctly categorized in each category. The results of this testing demonstrate that a neural network based classifier outperforms a rule-based classifier in terms of overall accuracy in this scenario.

Publication Date

5-1-2017

Language

English

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.