The Effects of Signal-to-Noise Ratio on Generative Adversarial Networks Applied to Marine Bioacoustic Data
Preprint
Abstract
In recent years generative adversarial networks (GANs) have been used to supplement datasets within the field of marine bioacoustics. This is driven by factors such as the cost to collect data, data sparsity and aid preprocessing. One notable challenge with marine bioacoustic data is the low signal-to-noise ratio (SNR) posing difficulty when applying deep learning techniques such as GANs. This work investigates the effect SNR has on the audio-based GAN performance and examines three different evaluation methodologies for GAN performance, yielding interesting results on the effects of SNR on GANs, specifically WaveGAN.Keywords
All Related Versions
This publication has 6 references indexed in Scilit:
- Low-Resource Generation Method for Few-Shot Dolphin Whistle Signal Based on Generative Adversarial NetworkJournal of Marine Science and Engineering, 2023
- Learning Stage-Wise GANs for Whistle Extraction in Time-Frequency SpectrogramsIEEE Transactions on Multimedia, 2023
- Dolphin vocal sound generation via deep WaveGANJournal of Electronic Science and Technology, 2022
- Deep whistle contour: Recall-guided learning from synthesisThe Journal of the Acoustical Society of America, 2019
- Signature whistles in free‐ranging populations of Indo‐Pacific bottlenose dolphins, Tursiops aduncusMarine Mammal Science, 2013
- Whistle characteristics in free-ranging bottlenose dolphins (Tursiops truncatus) in the Mediterranean Sea: Influence of behaviourMammalian Biology, 2011