Report of experience - invalid MusicXML file

[grarddeg]

After a few days of experience with Audiveris, I find it works reasonably well both on a vectorial pdf (generated by a music engraving software) and on a (good resolution) bitmap pdf (scanned printed score).
I did however encounter a problem with one of the scanned pages, the resulting MusicXML file could not be imported either by MuseScore (invalid MusicXML file) or by musicxml2ly. I enclose the zipped mxl file and the musicxml2ly log mxl file. I can also provide the uncompressed MusicXML file if need be.

p2.mxl.zip
musicxml2ly log

---

Démarrage musicxml2ly...
musicxml2ly: Reading MusicXML from /Users/grarddeg/Library/AudiverisLtd/audiveris/data/Krommer/p2.mxl ...
musicxml2ly: Input file /Users/grarddeg/Library/AudiverisLtd/audiveris/data/Krommer/p2.mxl is compressed, extracting raw MusicXML data
musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: warning: overfull measure? Expected: 269/4, Difference: -1/4
In:

In:

In:

musicxml2ly: Converting to LilyPond expressions...
musicxml2ly: warning: Negative skip -1/4 (from position 269/4 to 67)
musicxml2ly: warning: Negative skip -1/4 (from position 269/4 to 67)
musicxml2ly: warning: Negative skip -1/4 (from position 135/2 to 269/4)
musicxml2ly: Converting to LilyPond expressions...
Traceback (most recent call last):
File "/opt/local/bin/musicxml2ly", line 3482, in
main()
File "/opt/local/bin/musicxml2ly", line 3475, in main
voices = convert(filename, options)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/local/bin/musicxml2ly", line 3336, in convert
(voices, staff_info) = get_all_voices(parts)
^^^^^^^^^^^^^^^^^^^^^
File "/opt/local/bin/musicxml2ly", line 2923, in get_all_voices
voice = musicxml_voice_to_lily_voice(v)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/local/bin/musicxml2ly", line 2548, in musicxml_voice_to_lily_voice
main_event = n.to_lily_object(
^^^^^^^^^^^^^^^^^
File "/opt/local/share/lilypond/2.24.1/python/musicxml.py", line 892, in to_lily_object
event.duration = self.initialize_duration()
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/local/share/lilypond/2.24.1/python/musicxml.py", line 841, in initialize_duration
% (mxl_note.start, mxl_note._duration))
^^^^^^^^
NameError: name 'mxl_note' is not defined. Did you mean: 'Xml_node'?
Arrêté avec le code de retour 1.

[hbitteur]

When loading your .mxl file, MuseScore reports this error:

Fatal error: line 1779 column 18 Content of element duration does not match its type definition: Double content does not match the minExclusive facet..

Indeed a 0-duration rest is a very short rest! :-)

Could you post the input file (just the page which contains this measure 63), so that I can investigate further?

[grarddeg]

There you go. On measure 63, it didn't catch the 1-note d apparently.

Krommerp2.pdf

[hbitteur]

Yes, the double note was not recognized, though the related template seems to fit rather well. To be further checked.

Anyway, an empty measure should not export a 0-duration rest, this is a real bug. To be fixed.

[grarddeg]

As a follow-on, setting the duration to 48 wherever it is set to zero works, at least with musicxml2ly, the file gets imported into LilyPond. But I couldn't import it into MuseScore, maybe because both the issue below.
There is strange behaviour though, that I experienced also with the other scanned page : it creates as many voices (or parts) as there are staves in the original score and scatters the music between the voices/parts. This requires to copy each bit into a single voice and remove all others when it's done, which is a little tedious.