A compressed-domain audio fingerprint algorithm for resisting linear speed change
Liming Wu1, Wei Han1, 2, Songbin Zhou2, Xin Luo1, Yaohua Deng1
1School of Information Engineering, Guangdong University of Technology, Guangzhou 510006, China
2Guangdong Institute of Automation, Guangzhou 510070, China
Existing compressed-domain audio fingerprint algorithms have been able to be used to recognize audio information effectively according to hearing content, and are robust to common time-frequency domain distortion, including echo, noise, band-pass filtering, 32Kbps@MP3 and so on. However, they are poor in resisting linear speed change, which is a very common method for audio processing. In this paper, we propose a novel compressed-domain audio fingerprint algorithm. It is robust to large linear speed change via using auto-correlation function to reduce unaligned degree of MDCT spectrum sub-bands’ energy. Besides, it is similar with existing compression-domain audio fingerprint algorithms on the other aspects.