Segmentation and characterization of acoustic event spectrograms using singular value decomposition

作者：

Highlights：

• Acoustic event spectrograms are segmented using left singular vector.

• Segmented acoustic event spectrograms are characterized.

• Robustness of the proposed approach is evaluated in noisy conditions.

• Recognition performance is compared with other methods.

摘要

•Acoustic event spectrograms are segmented using left singular vector.•Segmented acoustic event spectrograms are characterized.•Robustness of the proposed approach is evaluated in noisy conditions.•Recognition performance is compared with other methods.

论文关键词：Acoustic Event Classification (AEC),Singular Value Decomposition (SVD),Singular vectors,Spectrogram segmentation,Spectrogram characterization

论文评审过程：Received 5 May 2018, Revised 25 October 2018, Accepted 2 December 2018, Available online 3 December 2018, Version of Record 6 December 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.12.004