Identify Research Gaps to narrow scope of investigation

[SteveAndCow]

Model Specialization: Lu et al. (2025) suggest designing different classification structures for different categories of music, indicating a gap in achieving a truly universal MSA model that performs optimally across diverse genres (e.g., pop vs. classical).

For example, since a few of us have interest in electronic music, and in initial tests of using SongFormer, the section labels aren't necessarily relevant enough to the genre -- i.e. labelling the buildup section of a DNB song as "prechorus, chorus".