An Efficient Music Search Method in Large Audio Database

In this paper we study the music content search method in a large audio database. First we study the pitch feature extraction algorithm, we propose a novel method on pitch estimation for music signals. Second, we study the template matching algorithm for large database, and we propose an effective method using dynamic template wrapping and linear scaling distance. Finally, experimental results are provided. The results show that the proposed system is efficient in audio data searching compared with traditional template matching method.