Two postgraduate students from the Department of Electrical & Electronic Engineering at Stellenbosch University have earned international recognition at one of the world’s leading audio machine learning competitions.

Christiaan Geldenhuys (PhD candidate) and Günther Tonitz (Master’s student) secured second place overall at BioDCASE 2025, the specialised bioacoustics track of the IEEE Detection and Classification of Acoustic Scenes and Events (DCASE) Challenge.

Organised by the IEEE Audio and Acoustic Signal Processing (AASP) community, DCASE is a globally recognised benchmark competition in machine learning for audio analysis. BioDCASE 2025 challenged teams to develop advanced algorithms capable of detecting Antarctic blue and fin whale calls in complex underwater recordings.

The dataset comprised seven call classes; however, evaluation was conducted at a collapsed level, requiring models to distinguish between three grouped classes. The task was particularly demanding due to low signal-to-noise ratios across multi-year Antarctic datasets, with whale calls occurring only approximately 5% of the time (5.1% event ratio).

This research supports long-term acoustic monitoring of whale populations recovering from near-extinction due to industrial whaling, demonstrating the powerful role of engineering and machine learning in marine conservation.

Competing against six international research teams, the Stellenbosch team achieved the second-best technical performance. At the BioDCASE workshop in Barcelona, Spain, their peer-reviewed paper received the Jury Award for Best Paper — notably ranked above the paper of the overall winning team.

Supervised by Prof Thomas Niesler, the team expanded on their work and later received another Best Paper Award at the Southern Africa Telecommunication Networks and Applications Conference (SATNAC) 2025.

These achievements highlight the strength of advanced signal processing and machine learning research within the Faculty of Engineering, and underscore Stellenbosch University’s growing international impact in audio machine learning and environmental monitoring.

Congratulations to Christiaan, Günther and Prof Niesler on this exceptional accomplishment.