Davel, MH.Versfeld, DJJ.Fourie, Edrich2024-08-072024-08-072024-04https://orcid.org/0000-0002-0150-1531http://hdl.handle.net/10394/42681Master of Engineering in Computer and Electronic Engineering, North-West University, Potchefstroom CampusPassive acoustic monitoring with hydrophones makes it possible to detect the presence of marine animals over large areas. For monitoring to be cost-effective, this process should be fully automated. We explore a new approach to detecting whale calls, using an endto- end neural architecture and traditional speech features. We compare the results of the new approach with a Convolutional Neural Network (CNN) applied to spectrograms, currently the standard approach to whale call detection. Experiments are conducted using the “Acoustic trends for the blue and fin whale library” from the Australian Antarctic Data Centre (AADC). We experiment with different types of speech features (Mel Frequency Cepstral Coefficients (MFCC) and Filter banks (Fbanks)) and different ways of framing the task. We demonstrate that a time-delay neural network is a viable solution for whale call detection, with the additional benefit that spectrogram tuning – required to obtain high-quality spectrograms in challenging acoustic conditions – is no longer necessary.enConvolutional neural networksTime-delay neural networksSpectrogramsMel frequency cepstral coefficientsFilter banksThesis