Automatic Detection of Flaws in Recorded Music using Wavelet Fingerprinting
Laney, Ryan
Laney, Ryan
Abstract
This thesis describes the application of wavelet fingerprinting as a technique to analyze and automatically detect flaws in recorded audio. Specifically, it focuses on time-localized errors in digitized wax cylinder recordings and contemporary digital media. By taking the continuous wavelet transform of various recordings, we created a two-dimensional binary display of audio data. After analyzing the images, we implemented an algorithm to automatically detect where a flaw occurs by comparing the image matrix against the matrix of a known flaw. We were able to use this technique to automatically detect time-localized clicks, pops, and crackles in both cylinders and digital recordings. We also found that while other extra- musical noises, such as coughing, did not leave a traceable mark on the fingerprint, they were distinguishable from samples without the error.
Description
Thesis is part of Honors ETD pilot project, 2008-2013. Migrated from Dspace in 2016.
Date
2011-05-04
Journal Title
Journal ISSN
Volume Title
Publisher
Collections
Research Projects
Organizational Units
Journal Issue
Keywords
Wavelet, Wavelet thumbprint, Wavelet fingerprint, Music, Recording, Wax cylinder, Phonograph, Non-destructive evaluation