Feature Extraction for Audio Fingerprinting Using Wavelet Transform

Abstract

This paper proposes a novel feature extraction scheme for audio fingerprinting using discrete wavelet transform (DWT). The proposed scheme reduces the granularity, i.e., the minimal length of audio, needed for identification in an audio fingerprinting system. The scheme first decomposes the video frame into sub-bands by DWT. Then, it calculates statistical values from the DWT coefficients and extracts features according to the statistical data. Finally, it uses the features to construct the fingerprint. The proposed scheme has two advantages: (1) it needs smaller fingerprint granularity than other previous work; (2) it is not only reliable but also robust against various signal degradations according to the experimental results.