A method for enhancement and automated extraction and tracing of Odontoceti whistle signals base on time-frequency spectrogram. (May 2021)

• Record Type: Journal Article
• Authors: Wang, Xianquan; Jiang, Jiajia; Duan, Fajie; Liang, Chunjiang; Li, Chunyue; Sun, Zhongbo; Lu, Ruichen; Li, Fangyi; Xu, Junyu; Fu, Xiao
• Is Part Of: Applied acoustics. Volume 176(2021)
• Publication Date: 2021-05
• Subjects: Underwater acoustics -- Extraction of Odontoceti whistle signal -- Time–frequency tracing -- Marine acoustics technology -- Underwater acoustics signal processing; Acoustical engineering -- Periodicals; Periodicals; 620.2
• DOI: 10.1016/j.apacoust.2020.107698
• Languages: English
• ISSNs: 0003-682X

Abstract

Abstract: The extraction, time–frequency (TF) tracking and enhancement of whistle signals play an important role in biological research or non-biological research about whistle signals of cetaceans. Underwater noise is a major hurdle in the automated extraction of whistle from sound recording files. Another major hurdle is that some whistle signals that overlap each other need to be separated automatically. This paper proposes a whistle signal processing algorithm based on time–frequency spectrum (TFS) image and morphology processing, which can extract whistle signals from noisy background noise and achieve TF tracking and enhancement of whistle signals. This algorithm has revealed perform well during dealing with the above-mentioned challenging problem. Experiments show that, under low signal-to-noise ratio (SNR) conditions, the algorithm can still accurately extract, and track, and enhance the whistle signals. In addition, the algorithm can also separate whistle signals which overlap each other.