Fingerprint generation of audio signal using Difference of Gaussian

In the recent times, several audio search sites use audio fingerprinting technology which provides the ability to identify short and unlabeled audio clips in a rapid and reliable manner. However, if the number of fingerprints per audio file increases, the amount of query data for audio retrieval also increases. In this paper, therefore, a novel method is proposed that can reduce the number of fingerprints by using Difference of Gaussian and provide a degree of performance as same as that of the existing techniques. The proposed method uses Difference of Gaussian for feature extraction so that it is used effectively for audio retrieval even if the quality of the audio signal has been changed.