An indexing and search strategy for fingerprint databases

An audio fingerprint is a content-based compact signature that summarizes an audio recording. A song can be recognized by matching an extracted fingerprint to a database of known fingerprints. Audio fingerprinting must solve the two key problems of fingerprint extraction and database search. In this paper, we are given a fingerprint database of songs and focus our attention on the problem of efficient database search. The high dimensionality of fingerprints suffers from the curse of dimensionality, i.e., as the dimension increases, the search performance decreases exponentionally. We describe a new strategy for indexing and search for fingerprint databases and then present experimental results for a database of 2,000 songs containing about 4,000,000 sub-fingerprints. The preliminary result of our experiments shows encouraging performance.