Machine Learning for Music Processing

Audio Fingerprinting

Title

Audio Fingerprinting Based on Normalized Spectral Subband Moments

Author

Jin S. Seo, Minho Jin, Sunil Lee, Dalwon Jang, Seungjae Lee, and Chang D. Yoo

Abstract

The performance of a fingerprinting system, which is often measured in terms of reliability and robustness, is directly related to the features that the system uses. In this letter, we present a new audio-fingerprinting method based on the normalized spectral subband moments. A threshold used to reliably determine a fingerprint match is obtained by modeling the features as a stationary process. The robustness of the normalized moments was evaluated experimentally and compared with that of the spectral flatness measure. Among the considered subband features, the first-order normalized moment showed the best performance for fingerprinting.

audio_fingerprint_1.jpg

Overview of (a) audio fingerprinting system used for identification and
(b) fingerprint extraction from spectral subband moments

audio_fingerprint_2.jpg

ROC curves for four sets of distortions; N = 27, N = 54.
(a) Distortion set 1. (b) Distortion set 2. (c) Distortion set 3. (d) Distortion set 4.

 

Related Papers

1. Jin S. Seo, Minho Jin, Sunil Lee, Dalwon Jang, and Seungjae Lee and Chang D. Yoo, "Audio fingerprinting based on normalized spectral subband moments," IEEE Signal Processing Letters, vol. 13, no. 4, pp. 209-212, April 2006. 

2. Minho Jin and Chang D. Yoo, "Temporal Dynamics for Spectral Sub-band Centroid Audio Fingerprints," ICME2007.

3. Sungwoong Kim and Chang D. Yoo "Boosted binary audio fingerprint based on spectral subband moments," ICASSP2007, honolulu, USA, April 15-20, 2007.

4. Jin S. Seo, Minho Jin, Sunil Lee, Dalwon Jang, and Seungjae Lee and Chang D. Yoo, "Audio fingerprinting based on normalized spectral subband centroids," ICASSP2005, Philadelphia, USA, March 18-23, 2005.