An Efficient Hash Count Indexing and Searching Scheme for Audio Fingerprinting

Simarjeet Singh Bhatia, Rupali Bhartiya

Abstract: In todays dynamically changing world music databases are the one which are updating at a great speed and so is the identification process. In order to recognize a song rapidly and accurately we use an audio fingerprint. An audio fingerprint summarizes an audio recording or a song. A song can be recognized or identified by matching its fingerprint to the fingerprints present in a database. It has a wide variety of applications in broadcast monitoring, connected audio (FM/AM), music library and many more. This process of audio fingerprinting is divided into two phases namely fingerprint extraction and fingerprint matching. In this paper focus on fingerprint matching phase has been kept. We propose a new method of indexing and search algorithm. This indexing method will provide a base for our search algorithm to make it hybrid and more efficient. Theoretically and practically our algorithm provides better results as compared to Haitsma and Kalker method.

Keywords: sub-fingerprint, audio fingerprint, hashing, bunching, indexing