The workflow with importing/exporting MusicXML is really great—it opens up a whole new way of working.

Congratulations and thanks to Janusz and the OPMO team!

Greetings

André

I totally agree. The import of MusicXML is looking really reliable (only one glitch so far with a file) and the opportunities opened up by this new functionality are fantastic. Building libraries of material from xml scores rather than having to input OMN script manually is going to be a big time saver for me, but I’m already enjoying the flexibility of this new way of working and looking forward to the creative potential of working across platforms. So pleased with this new (and quite unexpected!) update - so thanks to Janusz and everyone involved with this development!

bw

Andrew

Some fixes to the notation display are on the way and will be available soon.

Converting from MusicXML to OMN was quite a task, especially when dealing with the many errors in the MusicXML script.

And this is part of the reason why...

https://guides.loc.gov/music-notation-preferred-preservation-formats-for-digital-scores/musicxml

A few more observations about the use of MusicXML extraction functions, as having now had a lot longer to work with them. This new functionality has been a massive improvement for my workflow. I like to use a lot of material in libraries to draw on quickly. Previously, I had to tediously enter much of this material manually. Using the XML extraction takes almost all of the manual entry away, Instead, the manual work is just tidying up errors of extraction or to tidy up files used to extract. I have found the OM functions to be - mostly - extremely effective and accurate. I have noticed the following:

a) using XML files from the Musescore repository - which I thought would be a main place for obtaining XML scores - has been very disappointing. The majority of the files I have obtained from Musescore will not convert due to errors. That’s not an OM problem, but quality of the files on Musescore, which I presume is because they are user uploaded and of low quality. I’ve found very few convert properly of those I have on file already or downloaded specifically to use with OMN extraction. There is no improvement in checking them in Dorico or Musescore first and resaving. They still do not work properly.

b) extremely good quality of XML extraction when creating the XML files myself from various sources

c) in particular, I have had - much to my surprise - an excellent success rate when using optical recognition software to convert PDFs to XML rather than using XML files generated by others - the XML creation element of this software must be very close to XML standard with good error checking because I have found very few errors at all in OM when converting these files

d) I have found that often the musicxml-to-score function throws an error on an XML file but using musicxml-to-omn does not - and the voice extraction with musicxml-to-omn works probably 95% of the time with the files I’ve used. Again, those that don’t work are problem with the files and not with OM.

After trying to establish a good workflow, I have actually found that I prefer the xml-to-omn function with the individual voice extraction key for most tasks - so d) is fine for what I need. Writing some functions to combine the extraction of individual voices into various combinations to produce usable OMN material for immediate use is a simple matter and quick to set up.

For anyone looking at OMR conversion - which gives total freedom to create XML files yourself from PDFs or scanned scores - I have found that both Newzik and Soundslice are very accurate compared with older options. Newzik’s online machine learning system is worth the very reasonable subscription rate and the advantage over the similar SoundSlice is that there is no limit on how many pages can be scanned and converted. I also find Soundslice to be slower to use because of the final manual checking the system does, which for a longer scan can take a long time. With both options, there is still some manual clean up of the scan before converting to XML, but this work is much reduced on manual work required to do enter everything previously, so a win-win solution for my way of working.

Hope that’s helpful to some

bw

Andrew

Thank you for your observations.

I will publish a function that could be useful for retrieving complete information about the XML file (score).

When extracting the instruments with musicxml-to-omn, you will be able to access the time signature, tempo, and other details, which you can then add to your score file.

(musicxml-metadata "Webern-Variationen-Für-Klavier.xml")
=> (:title "Variationen für Klavier" :movement-title "I" :composer "Anton Webern" :lyricist nil
    :date "2024-03-04" :software "MuseScore 2.3.2" :source nil :copyright "UE 10881"
    :relation nil :key-signature ((c major 143)) :time-signature ((3 16 54) (2 4 23) (3 2 66))
    :tempo ((e. 40 54) (q 160 23) (h 80 66)) :clef-layouts (treble-layout treble-layout)
    :layout-name ("Piano" "") :layout-abbr ("Pno." "") :instruments (inst1 inst2)
    :channel (1 1) :program (1 1) :volume (100 100) :pan (0 0))

musicxml-metadata function was implemented in Opusmodus version 4.0.30025

That’s a really useful addition - thank you!