AN EXPLORATION OF MUSIC STRUCTURE SEGMENTATION USING EEG DATA AND MSAF ALGORITHMS
Neha Rajagopalan (Stanford University)*, Blair Kaneshiro (Stanford University)
This paper will be presented in person
Inter-subject correlation (ISC) of electroencephalographic (EEG) responses to natural stimuli is thought to index at- tention, engagement, and salient stimulus events. We in- vestigate correspondences between EEG ISC and com- putationally derived music structure segmentation bound- aries. Using a publicly available EEG dataset, we com- puted time-resolved ISC of spatially optimized responses to a full-length Bollywood song. We also used the MSAF toolbox to compute structural segmentation boundaries from the stimulus audio. Preliminary results reveal that dif- ferent segmentation algorithms produce different boundary timestamps and that some but not all segmentation bound- aries align with ISC peaks. These findings encourage fur- ther research aimed toward advancing music neuroscience and multimodal approaches to music structure analysis