What MusicXML is: the score as data

It has happened to me more times than I would like: I finish writing a score in one program, try to open it in another, and it arrives broken — slurs are lost, the song lyrics, sometimes even a note here and there. That very concrete problem, moving a score from one place to another without it falling apart, is exactly the one MusicXML came to solve.

What MusicXML is

MusicXML is an open format for exchanging digital scores. It is maintained by the W3C Music Notation Community Group, its current version is 4.0 (2021), and today more than 250 notation, editing, and music analysis programs support it. In short: it is the standard format that lets two different applications hand a score to each other and understand one another.

Notation, not sound

It is worth not confusing it with MIDI. MIDI describes a performance — which note sounds, when, and how hard; it is the plane of execution. MusicXML describes the written score: the clef, the key signature, the rests, the slurs, the articulations, the lyrics under the notes. One stores how something is played; the other, how it is written.

They are two complementary views of the same music. That is why many programs export to both: to MIDI so it sounds, to MusicXML so it can be printed, edited, or analysed.

What it looks like inside

The “XML” in the name is not decorative: a MusicXML file is plain text, readable by a person and by a machine. A single note — a middle C lasting a quarter — is described like this:

<note>
  <pitch>
    <step>C</step>
    <octave>4</octave>
  </pitch>
  <duration>4</duration>
  <type>quarter</type>
</note>

You do not need to understand every tag to grasp the idea: each musical element — pitch, octave, duration, figure — is written down as structured data. That is what makes the score stop being an image and become manipulable information.

MusicXML versus MIDI

Neither replaces the other. If you want the music to sound, MIDI; if you want it read and preserved as a score, MusicXML.

Why it matters in research

For anyone who studies music with computational tools, MusicXML is a bridge. It lets a score travel between programs without losing information, be archived in an open format that does not depend on a single vendor, and be treated as data for analysis. It is also one of the natural outputs of automatic music transcription: when a system listens to a recording and infers the notes, MusicXML is one of the formats in which it can write the result so that a person can read it, correct it, and reuse it. That same readability is what makes it useful in Music Information Retrieval, where analysis often starts from the symbolic plane rather than from the audio.

Why it matters to me

I come back to the idea that runs through everything I do: joining instruments with systems that know how to make use of what happens when you play. A recording captures the sound; a score in MusicXML captures the structure, and it does so in a format that will still open many years from now. To document and study a repertoire — especially one that is poorly collected — having the music as open, readable data is not a luxury: it is the difference between preserving and losing.

References

The references this article draws on, and where to read further:

• W3C Music Notation Community Group. (2021). MusicXML 4.0.
• MusicXML — the format’s official site (documentation and tutorial).
• Good, M. (2001). MusicXML: An Internet-Friendly Format for Sheet Music. In Proceedings of XML 2001.
• Selfridge-Field, E. (Ed.). (1997). Beyond MIDI: The Handbook of Musical Codes. MIT Press.