Candidate Music IR Test Collections: A List

Donald Byrd

School of Informatics and School of Music, Indiana University

rev. mid-January 2006; partly rev. early Oct. 2009

 

By themselves, these collections might well be useful for a number of research projects. But with suitable queries and relevance judgments (if at all possible, human-produced), they would be much more useful, especially for cross-project comparisons via MIREX. NB: appearance in this list means only that I think the collection exists in machine-readable form somewhere and might be available; there are serious copyright as well as other availability issues for most of these! Exception: collections whose name is prefixed with "*" don't even exist yet in electronic form, as far as I know.

The RWC database is the one most clearly available for research purposes, and it has the valuable feature of containing the same music in both MIDI and audio form; but it's very small. The Uitdenbogerd & Zobel collection has the only existing set of human relevance judgments I know of, but the judgments are not at all extensive.

This list is a continuing work-in-progress: I attempt to maintain it as a public service, but I have very little time for it, and it's certainly not complete or up-to-date. The entries for music in symbolic-notation (CMN) form are probably the most complete. Those for CMN in image format—scanned scores and/or sheet music—are just a smattering of what’s available now; see the entry for the MLA Sheet Music Collections site, which has much more (links to about 60 collections, though most do not seem to have been digitized). The listings for audio are also very incomplete. Georg Holzmann's survey of music-IR datasets in different representations and for various purposes might also be of interest. In any case, comments on the current list are welcome.

Entries are in alphabetical order by name.

Name

Representation

Encoding

Complexity

Approx size

Comments

Bach Chorales

Event

MIDI (SMF)

Polyphonic

400 W

Short 4-part contrapuntal pieces: 185 (from BG v.39) + c.200 (from elsewhere)

Barlow & Morgenstern instrumental

CMN

??

Monophonic (>99%)

10,000 T

Themes of classical instrumental pieces plus (in the original book) an "index". NB: a few involve 2-3 voices. Cf. Themefinder.

Barlow & Morgenstern instrumental

Event

MIDI (SMF)

Monophonic (>99%)

10,000 T

Themes of classical instrumental pieces plus (in the original book) an "index". NB: a few involve 2-3 voices. See www.multimedialibrary.com/Barlow/index.asp.

Barlow & Morgenstern vocal

CMN

??

Monophonic

8,000 T

Themes of classical vocal pieces plus (in the original book) an "index".

CCARH MuseData

CMN

MuseData, Humdrum/kern

Polyphonic

4000 M

Complete movements of 881 Classical pieces from c. 1680 to 1815, at least 1/3 Bach. Includes 185 Bach chorales. See www.musedata.org

CD Sheet Music

"CMN"

Image

Polyphonic

7000?

Scanned scores. ca. 60 CDs, each of ca. 750-3800 pages of PDFs; total ca. 124,000 pages. Also available online from Byron Hoyt. See www.cdsheetmusic.com

Choral Public Domain Library

"CMN", CMN

Image; some Finale, etc.

Polyphonic

10,700

Contributed scores—all scanned, some also encoded—of choral works: www.cpdl.org . Site has many links to other free sheet-music sites.

Classical Archives

Event

MIDI (SMF)

Polyphonic

17,600

Contributed MIDI files of public-domain works; quality uneven. They claim a total of over 40,000 files (MIDI and audio) by over 2,000 composers. MIDI See www.classicalarchives.com

Classical Archives

Audio

Lossy (MP3, WMA)

Polyphonic

13,200

Recordings of public-domain works. They claim a total of over 40,000 files (MIDI and audio) by over 2,000 composers.

Classical MIDI Connection

Event

MIDI (SMF)

Polyphonic

??

Contributed MIDI files of public-domain works; quality uneven. See www.classicalmidiconnection.com/cmc

DML Digital Music Library

"CMN"

Image

Polyphonic

3500 W

Scanned scores; "more than 100,000 sheets of music". See www.digitalmusiclibrary.com

*ECOLM

CMN

LTN

Polyphonic

1000

Electronic Corpus of Lute Music: complete lute pieces

FreeScores.com

"CMN", event

Image, MIDI

Polyphonic

7500

See http://www.free-scores.com/

HymnQuest

CMN

??

Monophonic

13,300 T

Published by Stainer & Bell on CDROM (with content-based searching). See www.stainer.co.uk/hymnquest

International Music Score Library Project (IMSLP)

"CMN"

Image

Polyphonic

18,000 W

Public domain scores, and scores contributed by their composers. See http://imslp.org/wiki/Main_Page. (The collection was unavailable for some time due to threats of legal action, but the organizers seem to have taken great care to avoid future problems of that type.)

IRCAM

Audio

All both lossy (MP2) and lossless (CD)

Polyphonic

12,000 W

contemporary music works, all available with metadata; see www.musiquecontemporaine.fr

JHU/Levy Sheet Music

"CMN", CMN

Image; some Guido

Polyphonic

29,000 W=M

scanned sheet music (c.100,000 pages, c. 80% public-domain). See levysheetmusic.mse.jhu.edu

KernScores

"CMN", CMN

Humdrum/kern

Polyphonic

20,000 M

Includes a wide variety of music, with thousands of folk songs from many cultures as well as thousands of classical pieces from Gregorian chant to Medieval music to Bach (chorales, Brandenburg Concerti, and keyboard music) and Vivaldi, to Beethoven and Chopin to Prokofiev, plus smaller numbers of piano rags, barbershop quartets, harmony examples, etc. The site also converts to other symbolic encodings like GMN, MusicXML, MIDI file, etc. See kern.humdrum.net/help/tour

L of C/Copyright Sheet Music

"CMN"

Image

Polyphonic

62,500 W=M

scanned sheet music ("Music for the Nation: American Sheet Music": pieces copyrighted in 1820-1860 and 1870-1885). See memory.loc.gov/ammem/mussmhtml/mussmhome.html

L of C/Duke Sheet Music

"CMN"

Image

Polyphonic

3000 W

scanned sheet music ("Historic American Sheet Music, 1850-1920": pieces from the Duke collection)

Meldex (NZDML) Folksongs

CMN

Meldex

Monophonic

10,000 W=M

9500 German, Chinese, and Anglo-American folksongs from Schaffrath's collection and Digital Tradition. (Better would be *"MELDEX Plus": add back in the c. 200 containing tuplets NZDML removed.) See www.nzdl.org/musiclib

Musicnotes

CMN

Proprietary

Polyphonic

45,000 M

Many genres. See www.musicnotes.com

Musopen

Audio, "CMN"

??, Image

??

???

Scanned scores and recordings of classical music, with a simple way to view the score as you listen. Everything is public domain. Users can say how much they'd pay for a recording to be added; the website operator "will combine your donation with others who want the same piece then find a professional to record and donate the work". See www.musopen.com

Mutopia

CMN

LilyPond, image

Polyphonic

1300

Contributed classical works: www.mutopiaproject.org/index.html

NZDML Fake book

CMN

Meldex(?)

Monophonic??

1200

popular tunes

NZDML MidiMax

Event

MIDI (SMF)

Polyphonic??

100,000

Standard MIDI files (collected from the Web?)

Parsons

Pitch Contour

Pitch Contour

Monophonic

11,000 T

up/down/repeat encoding of pitches of themes of classical pieces

RISM

CMN

Plaine &Easie

Monophonic w/some chords

730,000 T

Incipits from works by over 19,500 composers. See www.nisc.com/Music/

RWC

Event

MIDI (SMF)

Polyphonic

300 W

Pieces in five groups: 100 pop, 15 "royalty-free" (sic), 50 classical, 50 jazz, 100 genres; most original, all freely available for research (same music as below). (Also includes a database of instrument sounds.) See staff.aist.go.jp/m.goto/RWC-MDB

RWC

Audio

Lossless (CD)

Polyphonic

300 W

Pieces in five groups: 100 pop, 15 "royalty-free" (sic), 50 classical, 50 jazz, 100 genres; most original, all freely available for research (same music as above). (Also includes a database of instrument sounds.) See staff.aist.go.jp/m.goto/RWC-MDB

Sheet Music Archive

"CMN"

Image

Polyphonic

??

"Downloadable editions of public domain classical music". About 80 composers are listed. See www.sheetmusicarchive.net

SheetMusicNow

CMN

Proprietary

Polyphonic

32,000 M

Classical and jazz. See www.sheetmusicnow.com

Sunhawk

CMN

Proprietary

Polyphonic

50,000 M

Many genres. See www.sunhawk.com

Templeton

"CMN"

Image

Polyphonic

22,000

scanned sheet music; at Mississippi State University

Themefinder

CMN

Humdrum/kern

Monophonic

37,000 T

Themes and incipits of classical pieces and folksongs. Its classical repertoire includes essentially all the Barlow & Morgenstern instrumental themes. See www.themefinder.org

UCLA Pop. American

"CMN"

Image

Polyphonic??

600

Pieces of popular American music

Uitdenbogerd & Zobel

Event

MIDI (SMF)

Polyphonic

10,500

MIDI files (collected from the Web). Some "real" relevance judgments exist.

Variations

"CMN"

Image

Polyphonic

300 W

Scanned scores; includes c. 70 complete operas. See www.dlib.indiana.edu/variations/scores

Variations2

"CMN"

Image

Polyphonic

??

Scanned scores

Variations2

Audio

Lossy (MP3)

Polyphonic

??

c. 9,000 hours

Wikifonia

"CMN"

MusicXML

Monophonic

1300 W?

"Wikifonia is the place for musicians to publish and collaborate on lead sheets." See www.wikifonia.org

Other possibilities:

Various

"CMN"

Image

Polyphonic

Music Library Association Sheet Music Collections web site: www.lib.duke.edu/music/sheetmusic/collections.html

 

As used here, "Polyphonic" means any music that’s not monophonic, i.e., that has more than one note sounding at a time, regardless of rhythmic independence. "Approx. size" is number of documents (for classical pieces, often individual movements).

CMN = some form of Conventional Music Notation, perhaps very incompletely encoded.

"CMN" = Conventional Music Notation in image form, not encoded.

image = GIF, TIFF, JPEG, etc. (These would be far more useful if encoded, e.g., via Optical Music Recognition.)

LTN = Lute Tablature Notation.

SMF = Standard MIDI File (the standard encoding of time-stamped MIDI).

TSM = some form of time-stamped MIDI.

Suffixes on "Approx size" are W = works, M = movements, T = themes or incipits. "W=M" means that all or virtually all works in the collection include just one movement.

Thanks to Jeremy Pickens, Craig Sapp, Rainer Typke, and Masataka Goto for their help with this document.