Reads ABC notation written one system of music at a time, and writes the systems out as one MusicXML score.

Each system is a header and one line per voice, the subset of ABC that the rokot-omr-2b music recognition model writes. A voice named by a staff number, such as V:1, is a part, and a voice named after a staff, such as V:2b, is a further voice of that staff's part. Slurs, grace notes, annotations and decorations other than dynamics are left out, as mx plays none of them.