Underwater Acoustic Signal Noise Reduction Based on a Fully Convolutional Encoder-Decoder Neural Network

• Published in: Journal of Ocean University of China Vol. 22; no. 6; pp. 1487 - 1496
• Main Authors: Song, Yongqiang, Chu, Qian, Liu, Feng, Wang, Tao, Shen, Tongsheng
• Format: Journal Article
• Language: English
• Published: Heidelberg Science Press 01.12.2023; Springer Nature B.V
• Subjects: Acoustic mapping; Acoustic noise; Acoustics; Audio data; Convolution; Deep learning; Earth and Environmental Science; Earth Sciences; Encoders-Decoders; Feature extraction; Frequency analysis; Machine learning; Marine environment; Meteorology; Neural networks; Noise reduction; Oceanography; Sonar detection; Sound detecting and ranging; Spectrograms; Target masking; Underwater; Underwater acoustics; Waveforms; Wavelet analysis; wavelet low-frequency analysis recording spectrogram; deep learning; convolutional encoder-decoder neural network
• ISSN: 1672-5182; 1993-5021; 1672-5174
• DOI: 10.1007/s11802-023-5458-z

Noise reduction analysis of signals is essential for modern underwater acoustic detection systems. The traditional noise reduction techniques gradually lose efficacy because the target signal is masked by biological and natural noise in the marine environment. The feature extraction method combining time-frequency spectrograms and deep learning can effectively achieve the separation of noise and target signals. A fully convolutional encoder-decoder neural network (FCEDN) is proposed to address the issue of noise reduction in underwater acoustic signals. The time-domain waveform map of underwater acoustic signals is converted into a wavelet low-frequency analysis recording spectrogram during the denoising process to preserve as many underwater acoustic signal characteristics as possible. The FCEDN is built to learn the spectrogram mapping between noise and target signals that can be learned at each time level. The transposed convolution transforms are introduced, which can transform the spectrogram features of the signals into listenable audio files. After evaluating the systems on the ShipsEar Dataset, the proposed method can increase SNR and SI-SNR by 10.02 and 9.5dB, respectively.