This release v2.0 adds note alignments to the ASAP dataset v1.1 and so creates the (n)ASAP dataset.
The note alignments where created on the basis of the beat annotations in ASAP and an anchor point-based note alignment algorithm available in a separate library parangonar
The note alignments are available in three formats:
tsv
files with note IDs and minimal time and pitch informationmatch
files containing linked notes and encoding a bare-bones score as well as the MIDI performance.match
can easily be parsed with partituraparangonada
files: a directory of five files used for visualization and editing with parangonada
The musicxml
score files have been updated with note IDs which can also easily be parsed with partitura. Other than the IDs the musicxml
files and all other unchanged files are taken from version 1.1 of the ASAP dataset.
All note alignments were visually inspected and those deemed without apparent errors are tagged as robust_note_alignment
in the metadata.csv
file.