An Asymmetric Matching Method for a Robust Binary Audio Fingerprinting

Audio fingerprinting has been successfully applied for the problems associated with the protection of intellectual property, management of large database and indexation of content. For a reliable fingerprinting system, improving fingerprint matching accuracy is crucial. In this letter, we try to improve a binary audio fingerprint matching performance by utilizing auxiliary information which is discarded while extracting fingerprint bits. Since it is not feasible to store all the auxiliary information in the fingerprint DB, we only utilize the auxiliary information obtained while extracting fingerprints from the input unknown audio. We call the proposed method an asymmetric fingerprint matching, since the auxiliary information is available only in one side. Experimental results show that the proposed asymmetric matching method is effective in improving fingerprint matching performance over the conventional Hamming distance.