Mismatch midi and musicxml scores

[DavidGarfinkle]

indexers.notes does not return the same data running on a midi file versus the musicxml output of that midi file as output by music21.converter.parse(midi).write('xml')

[vigliensoni]

Create a minimal example that we can send to the Music21 developers.

…

On Thursday, June 27, 2019, David Garfinkle ***@***.***> wrote: indexers.notes does not return the same data running on a midi file versus the musicxml output of that midi file as output by music21.converter.parse(midi).write('xml') — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#8?email_source=notifications&email_token=AADYFRNMEZIKXEIMSECT3UTP4S5MHA5CNFSM4H336AI2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4G4CLAUA>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AADYFRPEIBMDCE52FZ6TAG3P4S5MHANCNFSM4H336AIQ> .

[DavidGarfinkle]

def first_mismatch(midi_path):
     s = music21.converter.parse(midi_path)
     npsmid = indexers.NotePointSet(s)
     o = s.write('xml')
     npsxml = indexers.NotePointSet(music21.converter.parse(o))
     print(next(i for i, (x, y) in enumerate(zip(npsmid, npsxml) if x != y))

with test mass testdata/palestrina_masses/mid/000000000011521_Missa-Ut-re-mi-fa-sol-la_Credo_Palestrina-Giovanni-Pierluigi-da_file2.mid

and found the first mismatch occurs at a tied note

i think this means the bug is on our end, but there remains some thinking..

[DavidGarfinkle]

This bug occurs because music21.converter.parse(midi_file) extended quarter length durations on tied notes, whereas music21.converter.parse(midi_stream.write('xml')) ignored ties in calculating note lengths.

we can arrive at a solution by calling stream.stripTies() on the music21 xml output, but this will prevent occurrences from being found beyond the first note of a tied group

[DavidGarfinkle]

todo: include the expected behaviour of finding occurrences through ties in the documentation

& ask peter

[DavidGarfinkle]

the current solution has been to ensure that the data we search over is the same from which we extract measures