A T5 transformer that tokenizes MIDI into text-like events and translates them into beat and downbeat tokens achieves state-of-the-art symbolic beat tracking across four datasets.
3.1 Datasets For the training and evaluation of the model, datasets con- taining synchronized MIDI and beat annotations are essen- tial
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SD 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Beat and Downbeat Tracking in Performance MIDI Using an End-to-End Transformer Architecture
A T5 transformer that tokenizes MIDI into text-like events and translates them into beat and downbeat tokens achieves state-of-the-art symbolic beat tracking across four datasets.