公开(公告)号：KR1020100115849A

公开(公告)日：2010-10-29

申请号：KR1020090034455

申请日：2009-04-21

Applicant: 서울대학교산학협력단

Inventor： 김남수 ,  윤환식 ,  리유

IPC: G10L21/06 ,  G10L19/02 ,  G06F17/30

Abstract: PURPOSE: An audio fingerprinting system based on a multiple hashing technique is provided to configure sub fingerprint by each coefficient, thereby providing sub fingerprint strong against noise. CONSTITUTION: An audio is divided by an overlapped frame(50). Frequency conversion like DFT(Discrete Fourier Transform) is applied to each frame(51). Sub fingerprint strong against noise is obtained using information obtained from L continuous frames(53). Sub fingerprint is obtained about each of K frequency coefficients(54).

Abstract translation: 目的：提供基于多重散列技术的音频指纹系统，通过每个系数配置子指纹，从而提供抗噪声的子指纹。 构成：音频被重叠的帧（50）划分。 诸如DFT（离散傅里叶变换）的频率转换被应用于每个帧（51）。 使用从L个连续帧获得的信息获得强烈抵抗噪声的子指纹（53）。 获得关于K个频率系数中的每一个的子指纹（54）。