Notes
Slide Show
Outline
1
Music Encoding Initiative
(MEI)
  • Perry Roland
  • University of Virginia
  • pdr4h@virginia.edu
  • http://www.lib.virginia.edu/digital/resndev/mei/


2
History of MEI
  • how it all started
  • work on XML music markup begun (2000), originally named Music Description Language (MDL)
  • activities – ISMIR (2000, 2003), Int'l Conf. on Musical Applications using XML (2002), MusicNetwork Notation Workshop (2003), OASIS Music Notation Technical Committee (2003), Online Chopin Variorum Edition meeting and white paper (2004)
  • current version is 1.6 beta (2005), version 1.7 beta (2006)
3
History of MEI
  • MEI covered in: Electronic Musician (2003), Experiencing Music Technology (2004), XSLT 2.0 (2005)
  • Regarding music markup schemes listed at coverpages.org:
  • "[M]any of these initiatives will bear little fruit [;however,] [t]here are at least two really serious contenders: the Music Encoding Initiative (MEI) and the Standard Music Description Language (SMDL).  -  Kay, p. 4
4
Why bother?
  • encoding
    • preservation of cultural heritage materials
    • content-based searching
    • analysis
    • re-use
  • interchange
    • provide interoperability among a large number of representations (current & future)
5

Approaches to music
encoding

  • audio / performance
  • notation / written artefact
  • selection depends on definition of "musical work"


6
Performance
  • events are not discreet
    • example: a note's pitch may change over time
  • contains ambiguities
    • homophones: C# - D♭
    • composer "tricks": reinforcement of selected overtones, obscured boundaries, unusual performance methods, ...
  • records a single performance, not generalize-able
  • intellectual property rights problems
7
Notation
  • represents a generic, "Ur" version
  • comprehensive - contains (almost) all info. required for printing, performance, and analysis
  • contains (more or less) discreet "chunks"
    • example: each note has a single pitch
  • provides generally-accepted terminology
  • enormous corpus of copyright-free sources
8
A Good Encoding Is ...
  • declarative – specifies what a thing 'is', not what it looks like
  • human readable (and understandable!) – accurate in its use of names for key concepts and consistent in its terminology
  • formal – verifiable
  • explicit – context independent
  • flexible – allows selective feature encoding
  • extensible – accommodates unknown uses and future development
9
A Good Interchange Standard Is ...
  • software independent
  • comprehensive – generalized
  • N*(N-1) vs. N*2 translators
10
Advanced Features
  • internal references and annotations
  • rich meta-data header
  • work-level text
  • arbitrary segmentation
  • multiple views
  • variant readings
  • alignment with external material



11
Data Sources
  • printed music (via OMR)
  • notation editor (via GUI or MIDI input)


12
Transformation to MEI
  • Chopin, Barcarolle, op. 60, pg. 9
  • SharpEye -> MusicXML
  • MusicXML -> MEI
13
MEI Representation
  • <measure n="1">
  •   <staff def="1">
  •     <layer def="1">
  •       <beam>
  •         <note id="n1" pname="c" oct="6" dur="16" stem.dir="down" />
  •         <note pname="a" acci="s" oct="5" dur="16" stem.dir="down" />
  •         <note pname="f" oct="5" dur="16" stem.dir="down" />
  •         <note pname="g" oct="5" dur="16" stem.dir="down" />
  •         <note pname="b" acci="s" oct="5" dur="16" stem.dir="down" />
  •         <note pname="a" oct="5" dur="16" stem.dir="down" />
  •       </beam>
  •       <beam>
  •          <note pname="f" oct="5" dur="16" stem.dir="down" />
  •          <note pname="d" acci="s" oct="5" dur="16" stem.dir="down" />
  •         <note pname="a" acci="s" oct="4" dur="16"  stem.dir="down" />


  • <note pname="c" oct="5" dur="16" stem.dir="down" />
  •  <note pname="e" acci="s" oct="5" dur="16" stem.dir="down" />
  •  <note pname="d" oct="5" dur="16" stem.dir="down" />
  • </beam>
  • <beam>
  •   <note pname="c" oct="6" dur="16" stem.dir="down" />
  •   <note pname="a" acci="s" oct="5" dur="16" stem.dir="down" />
  •   <note pname="f" oct="5" dur="16" stem.dir="down" />
  •   <note pname="g" oct="5" dur="16" stem.dir="down" />
  •   <note pname="b" acci="s" oct="5" dur="16" stem.dir="down" />
  •   <note pname="a" oct="5" dur="16" stem.dir="down" />
  • </beam>
  • <!-- last beat omitted -->
  • </layer>
  • </staff>
  •  <!-- staff 2 omitted -->
  •  <hairpin form="dim" staff="1" tstamp="1" dur="12.5" place="below" />
  •   <phrase staff="1" start="n1" end="d0e843" bulge="+" />
  •   <pedal staff="2" tstamp="1" dir="down" />
  •   <pedal staff="2" tstamp="12" dir="up" />
  •   <octave staff="1" place="above" tstamp="1" dis="8" dur="6.5" />
  •   <dir staff="1" place="above" tstamp="7">loco</dir>
  • </measure>
14
MEI Visualization
  • MEI -> Mup (Music Publisher)
  • handles layout
  • plain text
  • notation (PostScript) and MIDI output
  • low cost
  • macro programming language (extensible)
15
MEI Visualization
  • MEI -> MusicXML -> Notation editor
  • currently working on more robust "round trip" (MusicXML -> MEI -> MusicXML) stylesheets
16
Next Steps
  • testing on wider range of material, more review of the DTD
  • creation of direct-to-MEI plug-in
  • creation of a rudimentary MEI editor
  • formulation of an MEI workgroup
17
 
18
Music Encoding Initiative
  • Perry Roland
  • University of Virginia
  • pdr4h@virginia.edu
  • http://www.lib.virginia.edu/digital/resndev/mei/