Fingerprint extraction of audio signal using wavelet transform
M. D. Kamaladas, M. Maxina Dialin · 2013
Due to the massive growth in computer technology enormous amount of music files are shared in the internet. The storage devices of music files are also changed from cassette tapes to MP3 players, mobile phones, DVD players. In order to solve problems in searching the needed songs among mass audio information, digital audio fingerprinting is involved. Audio fingerprint is a content based compact signature, which summarizes the audio. Fingerprinting systems extracts the acoustic characteristics of the audio and store it as a fingerprint and store it in a database. When the unknown, short snippet of an audio given as query, it calculates its characteristics and compared it to the stored database. Using the fingerprint matching algorithms, the particular audio data was successfully identified even if the audio data has been distorted severely. However the size of the fingerprint per music file is increased, then the amount of songs with their fingerprints in the database is less. So, a novel method proposed to reduce the size of the fingerprints thereby increasing the amount of songs stored in the database. The proposed method uses the wavelet transform for extracting the features so it is effectively used for identification even if the audio quality of the audio has been changed and the amount of fingerprints generated also less.