CODA: Cascaded Online Discontinuity-Aware Alignment for Real-Time Image-Based Score Following

TL;DR AI
2 min readKey summary
Researchers introduced CODA, a real-time score-following system that aligns live audio with sheet images.
CODA uses cascaded predictions from system to bar to note, improving stability across multiple levels of music structure.
It also uses silence-based recovery to handle repeats, jumps, and other score discontinuities.
The approach improves accuracy under real-time constraints, which is valuable for accompaniment, practice, and performance analysis.
