复杂环境下基于听觉子带能量特征的鸟鸣声端点检测  

Endpoint detection of bird sound in complex environment based on auditory sub-band energy feature

作　　者：王静宇[1] 张纯[1] 许枫[1] WANG Jingyu;ZHANG Chun;XU Feng(Institutes of Acoustics,Chinese Academy of Science,Beijing 100190,China)

机构地区：[1]中国科学院声学研究所,北京100190

出　　处：《计算机应用》2022年第S01期310-315,共6页journal of Computer Applications

基　　金：国家重点研发计划项目(2017YFC1403501)。

摘　　要：为了检测野外复杂噪声环境中的鸟鸣声信号,提出一种基于人耳听觉特性的听觉子带能量特征鸟鸣声端点检测方法。利用反映人耳听觉特征的Mel频率尺度变换,将鸟鸣声信号在频域划分为24个子带(称为Mel子带),分析鸟鸣声信号的Mel子带能量分布特征,选取能量值最高的Mel子带能量作为特征量进行端点检测。通过仿真和野外实测数据对比了基于短时能量法的端点检测性能,结果表明Mel子带能量法在信噪比(SNR)为-10 dB条件下仍能检测到鸟鸣声信号,对风声、海浪声等海岛环境噪声也具有较强的抗干扰性能,性能优于短时能量法。In order to detect bird sound signals in complex noise environments in the field,a bird sound endpoint detection method based on the auditory sub-band energy characteristics of human ear was proposed.The Mel-frequency transform,which reflects the auditory characteristics of human ear,was used to divide the bird sound into 24 sub-bands(called Mel sub-band)in frequency domain,the features of Mel sub-band energy of bird sound were analyzed,and the sub-band with highest energy was selected as the feature variable for endpoint detection.Compared with the short-time energy algorithm,the simulation and experiment results show that the proposed method can still detect the bird sound signal under the-10 dB Signal-to-Noise Ratio(SNR),and has strong anti-interference performance for wind,wave and other island environmental noise,and the performance is better than that of the short-time energy method.

关 键 词：鸟鸣声 端点检测 Mel频率 听觉特征 子带能量 

分 类 号：TN911.72[电子电信—通信与信息系统]