西华大学学报(自然科学版)
西華大學學報(自然科學版)
서화대학학보(자연과학판)
JOURNAL OF XIHUA UNIVERSITY(NATURAL SCIENCE EDITION)
2014年
5期
10-15
,共6页
音频指纹%功率谱%鲁棒性
音頻指紋%功率譜%魯棒性
음빈지문%공솔보%로봉성
audio fingerprinting%power spectrum%robustness
针对现有算法在某些信号畸变情况下鲁棒性不理想的问题,修改了频带滤波器组模块的功率谱子带划分方式和二维滤波器模块的子带间时频差分方式,分别提出了基于频域和时域差分的2种音频指纹改进算法。实验结果表明,改进算法在低信噪比、线性速度改变等方面的鲁棒性均优于现有的算法。
針對現有算法在某些信號畸變情況下魯棒性不理想的問題,脩改瞭頻帶濾波器組模塊的功率譜子帶劃分方式和二維濾波器模塊的子帶間時頻差分方式,分彆提齣瞭基于頻域和時域差分的2種音頻指紋改進算法。實驗結果錶明,改進算法在低信譟比、線性速度改變等方麵的魯棒性均優于現有的算法。
침대현유산법재모사신호기변정황하로봉성불이상적문제,수개료빈대려파기조모괴적공솔보자대화분방식화이유려파기모괴적자대간시빈차분방식,분별제출료기우빈역화시역차분적2충음빈지문개진산법。실험결과표명,개진산법재저신조비、선성속도개변등방면적로봉성균우우현유적산법。
This paper is concerned with the analysis of audio fingerprint algorithm based on Fourier transform .In allusion to occur-ring problems of the existing algorithm in robustness , the authors modified the mode of energy division and difference between the sub-bands and presented an improved audio fingerprint algorithm which applied the difference in frequency and time domain .The pro-posed algorithm was analyzed subsequently , and experimental results showed that the proposed algorithms improved the robustness in several aspects such as low SNR and linear speed change than existing algorithm .