Automatic bioacoustics noise reduction method based on a deep feature loss network

• DOI: 10.1016/j.ecoinf.2024.102517
• Publication Date: 2024-02-21T17:07:15Z
• Authors: Chengyun Zhang; Kaiying He; Xinghui Gao; Yingying Guo
• Keywords: Bioacoustics; Feature (linguistics)

Acoustic sensors that collect acoustic data over extended periods and broad ranges are widely used in bioacoustics monitoring. However, open environments, collected using can be subject to interference from various real-world noises, thereby influencing the subsequent analysis processing of bioacoustic data. Existing noise reduction methods limited their application because low efficiency, unsuitability for non-stationary noise, generally unimproved signal-to-noise ratio (SNR) efficacy, considerable amounts residual noise. These limitations hinder effective recorded signals which extraneous overlaps with bird vocalizations. In this study, we propose a method based on deep feature loss network sounds. The has rapid denoising speed more effectively remove background field recording without distorting spectrum. effects proposed were compared those speech enhancement generative adversarial network, web real-time communications denoising, other methods. ability these different noises was evaluated spectrograms objective evaluation measures such as SNR perceptual quality (PESQ). experimental results revealed our obtain higher SNRs PESQ scores than methods, increasing by up 35.83 dB following denoising.