A deep learning-based data augmentation method for marine mammal call signals

Jiaming JiangWanlu ChengShengwen Gong^*Jingjing Wang^*

• Qingdao University of Science and Technology, Qingdao, Shandong Province, China

In marine ecology research, it is crucial to accurately identify the marine mammal species active in the target area during the current season, which helps researchers understand the behavioral patterns of different species and their ecological environment. However, the difficulty and high cost of collecting marine mammal calls, coupled with limited publicly available datasets, results in insufficient data for support, making it difficult to obtain accurate and reliable identification results.To address this problem, we propose MarGEN, a deep learning-based augmentation method for marine mammal call signal data. This method processed the call data into Mel spectrograms, then designed a self-attention conditional generative adversarial network to generate new samples of Mel spectrograms that were highly similar to the real data, and finally reconstructed them into call signals using WaveGlow. This study marks the first application of generative adversarial networks to the field of data augmentation of marine mammal call signals. Experimental results showed that the MarGEN method significantly increased the number and diversity of call signals, improving the model's recognition accuracy of marine mammal call signals by an average of 4.7%, effectively promoting marine ecology research.