Improving Sound Event Detection Metrics: Insights from DCASE 2020
This work addresses evaluation biases in SED for audio processing researchers, offering an incremental improvement to existing metrics.
The paper analyzed biases in sound event detection (SED) system rankings from DCASE 2020, showing that conventional metrics like event-based and segment-based criteria have flaws such as strictness dependency on event length and lack of precision, while the Polyphonic Sound Detection Score (PSDS) overcomes these by providing robust, operating-point-independent evaluations.
The ranking of sound event detection (SED) systems may be biased by assumptions inherent to evaluation criteria and to the choice of an operating point. This paper compares conventional event-based and segment-based criteria against the Polyphonic Sound Detection Score (PSDS)'s intersection-based criterion, over a selection of systems from DCASE 2020 Challenge Task 4. It shows that, by relying on collars , the conventional event-based criterion introduces different strictness levels depending on the length of the sound events, and that the segment-based criterion may lack precision and be application dependent. Alternatively, PSDS's intersection-based criterion overcomes the dependency of the evaluation on sound event duration and provides robustness to labelling subjectivity, by allowing valid detections of interrupted events. Furthermore, PSDS enhances the comparison of SED systems by measuring sound event modelling performance independently from the systems' operating points.