NotationAndEncoding

Music Notation and Encoding: 

Music Notation and Encoding ISMIR Graduate School, Barcelona 2004 Musicology 2 Frans Wiering, ICS, Utrecht University

Outline: 

Outline Common Music Notation Other notations Encoding music notation Music corpora

How to Read Music Without Really TryingDonald Byrd, School of Music, Indiana UniversityRev. August 2004 (adapted FW): 

How to Read Music Without Really Trying Donald Byrd, School of Music, Indiana University Rev. August 2004 (adapted FW) Four basic parameters of a definite-pitched musical note pitch: how high or low the sound is: perceptual analog of frequency duration: how long the note lasts loudness: perceptual analog of amplitude timbre or tone quality Above is decreasing order of importance for most Western music

CMN (Common/Conventional Music Notation)(after Don Byrd): 

CMN (Common/Conventional Music Notation) (after Don Byrd) CMN shows at least six aspects of music: NP1. Pitches (how high or low): on vertical axis NP2. Durations (how long): indicated by note/rest shapes NP3. Loudness: indicated by signs like p, mf, etc. NP4. Timbre (tone quality): indicated with words like “violin”, “pizzicato”, etc. Start times: on horizontal axis Voicing: mostly indicated by staff; in complex cases also shown by stem direction, beams, etc. Also shown: measures (beat groups) pitch modifiers (flats and sharps)

The keyboard (www.keyscreen.com): 

The keyboard (www.keyscreen.com)

Durations (American English)(library.thinkquest.org): 

Durations (American English) (library.thinkquest.org) Whole Half Quarter Eighth Sixteenth etc. duration tricks: rests:

Rhythm and meter: 

Rhythm and meter rhythm: the pattern of (relative) durations measure: recurring beat pattern meter: type of measure, expressed in fraction number of beats type of beat eg: 4/4, 7/2

Extensibility of notation (representational completeness): 

Extensibility of notation (representational completeness) CMN seems closed system (like an alphabet), but it isn’t historical change creativity precision (Mahler scores!) abbreviation example: chord abbreviations, as in fake books http://www.eddielandsberg.com/ jeanine_800x965.jpg Extremes of CMN collected by Don Byrd at http://php.indiana.edu/~donbyrd/ CMNExtremes.htm

Before CMN (1): 

Before CMN (1) neumes, c. 900 (from New Grove) plainchant, 17th c. (www.loc.gov) describe differences to CMN a little

Before CMN (2): 

Before CMN (2) mensural notation, early 16th c. (www.loc.gov) (from DIAMM)

Tablature (exx: New Grove): 

Tablature (exx: New Grove) 18th c. lute tablature koto tablature klavarskribo (20th c.)

Different aspects of notation: 

Different aspects of notation mnemonic neumes framework for improvisation fake book basso continuo John Cage performer instruction (lute) tablature much CMN composer’s intention 19th/20th century classical music

What is a basic musical ‘unit’: 

What is a basic musical ‘unit’ example units song the performance/recording classical music work: independent creation, realized in (more than one) performance and codified in notation early music instance: piece exists in many different forms (but see also Chopin first editions!) number of voices, levels of elaboration etc. Opera improvisation framework: archetypal melody, chord pattern, ‘raga’ consequences: the work ‘is’ not always the notation not all aspects of notation are always equally important (may reflect local circumstances rather than intention) it may be necessary to derive supplementary information but how?

Summary: 

Summary principal dimensions: pitch and time additional dimensions complexity non-CM notations relation to basic musical unit

Encoding music notation: 

Encoding music notation General characteristics Purposes of music encoding before MIR Surveys Requirements Some (once) popular encoding systems Unification and interchange Music corpora What if you want to work with encoded notation

General characteristics: 

General characteristics translating notation to digital representation or note-like information, as in MIDI alphanumeric and binary encoding systems very many formats have been proposed and are still in use interchange is a problem generally not designed for retrieval

Purposes of music encoding before MIR: 

Purposes of music encoding before MIR Music printing and publishing DARMS project and encoding (60s-80s) proprietary systems since the 80s (Finale, Sibelius) encoding systems tend to be exhaustive Cataloguing RISM project: cataloguing all music written before 1800. Beginnings in Plaine and easie code non-exhaustive systems Music analysis ESAC: folksong collections Humdrum: flexible encoding system for arbitrary (combination of) parameters Sound Control MIDI playing instructions

Surveys: 

Surveys www.music-notation.info Gerd Castan 71 different music encoding systems listed with links to software and web pages definitely incomplete Beyond MIDI. The Handbook of Musical Codes. ed. Eleanor Selfridge-Field (1997) c. 27 music encoding systems, detailed descriptions non-proprietory formats only

Requirements: 

Requirements some ideas representational completeness aptness extensibility efficient structure follows music mnemonic More on creating encoding systems: David Huron (1992) Beyond Midi (chapter by David Halperin)

Some (once) popular encoding systems: 

Some (once) popular encoding systems Plaine And Easie Code Barry S. Brook, Murray Gould (1964) writing music with a typewriter (card) catalogues DARMS Stefan Bauer-Mengelberg (1963); Raymond Erickson cheap printing of avant-garde music, very rich encoding set standard for later encoding systems later: analytical applications [right:] ''4CC/GG/AA/GG/FF/EE/D8.D6E/2C://: [left: ] ,4C'C/EC/FC/EC/D,B/'C,A/FG/2C://: !I1 !G,!F !M2:4 6Q 6Q / 10Q 10Q / 11Q 11Q / 10Q 10Q / 9Q 9Q / 8Q 8Q / 7Q 7E. 8S / 6H :/: !-50 K next line employs the 'carry feature' for durations $ !M2:4 4Q 11 / 13 11 / 14 11 / 13 11 / 12 10 / 11 9 / 7 8 / 4H :/:

More: 

More ESAC Helmut Schaffrath (1980s) analysis of folksong monophonic; phrase structure encoded. Used for testing automatic segmenting (Jane Singer?) MuseData Walter Hewlett, late 1980s logical content of musical scores printing, analysis, generation of sound right: KEY[C0000 16 C2/4] MEL[1__1__ 5__5__ 6__6__ 5__5__ 4__4__ 3__3__ 2__2_.3 1___] left: KEY[C0000 4 C2/4] MEL[--1-1 -3-1 -4-1 -3-1 -2--7 -1--6 --4--5 --1_] $ Q:4 K:0 T:2/4 S:2 C1:4 C2:22 S C5 4 1 q d1 C5 4 1 q d1 back 8 C3 4 2 q u2 C4 4 2 q d2 measure 2 G5 4 1 q d1 G5 4 1 q d1 back 8 E4 4 2 q d2 C4 4 2 q d2

Humdrum: 

Humdrum David Huron, c. 1990 http://csml.som.ohio-state.edu/Humdrum/ syntax for developing (light-weight) encoding systems data organised in parallel ‘spines’ humdrum representation for CMN: kern humdrum tools (UNIX) sample analytical problems (total: 350): Locate instances of the pitch sequence D-S-C-H in Shostakovich's music. Are German drinking songs more likely to be in triple meter. Determine whether Haydn tends to avoid V-IV progressions. Find all woodwind quintets in compound meters that contain a change of key. Classify cadences as either authentic, plagal or deceptive. Determine whether the words `high,' `hoch,' or `haut' tend to coincide with higher pitches in a vocal work. many can be rephrased as music IR problems **kern **kern *staff2 *staff1 =1- =1- *clefF4 *clefG2 *k[] *k[] *M2/4 *M2/4 *^ * 2ryy 4C 4cc . 4c 4cc =2 =2 =2 2ryy 4e 4gg . 4c 4gg =3 =3 =3 2ryy 4f 4aa . 4c 4aa =4 =4 =4 2ryy 4e 4gg . 4c 4gg =5 =5 =5 2ryy 4d 4ff . 4B 4ff =6 =6 =6 2ryy 4c 4ee . 4A 4ee =7 =7 =7

MIDI: 

MIDI controlling electronic instruments pitch represented by key number different types of events note on/off, and many others used for (very lossy) data exchange Not mentioned Guido Score MFile 1 2 1024 MTrk 0 TimeSig 2/4 24 8 0 KeySig 0 major 0 Tempo 500000 16385 Meta TrkEnd TrkEnd MTrk 0 Meta TrkName "Acoustic Grand Piano" 0 PrCh ch=1 p=0 0 On ch=1 n=72 v=64 0 On ch=1 n=48 v=64 1024 Off ch=1 n=72 v=0 1024 On ch=1 n=72 v=64 1024 Off ch=1 n=48 v=0 1024 On ch=1 n=60 v=64 2048 Off ch=1 n=72 v=0 2048 Off ch=1 n=60 v=0 2048 On ch=1 n=79 v=64 2048 On ch=1 n=64 v=64 3072 Off ch=1 n=79 v=0 3072 On ch=1 n=79 v=64 3072 Off ch=1 n=64 v=0 3072 On ch=1 n=60 v=64 4096 Off ch=1 n=79 v=0

Unification and interchange: 

Unification and interchange formats designed for interchange between applications rich descriptions of musical content SMDL ISO 10743. Based on HyTime/SGML domains logical (cantus, the abstract work); markup defined visual (score); container gestural (performance); container analytical; container never used; concepts circulate NIFF interchange of music notation data, page-oriented was supported by part of the industry still alive???

MusicXML: 

MusicXML developed by Michael Good (www.recordare.com) quotes from the site: open format, usable by as many applications as possible Dolet software uses MusicXML to provide a "universal translator" between music notation programs.

Music corpora (1): 

Music corpora (1) RISM (PAEC) 470.000 incipits on CD-ROM MIDI (everywhere) http://www.classicalarchives.com/ (8000 works) restricted access uneven quality CCARH http://www.ccarh.org Musedata, Humdrum (nearly 4000 movements of classical works) sources: out-of copyricht editions of Great Classical Composers quality of encoding

Music corpora (2): 

Music corpora (2) Graphic formats Choral Public Domain Library (http://www.cpdl.org/) 7800 scores sheet music, mostly in graphic formats Esac http://www.esac-data.org/ 20.000 songs and instrumental melodies, mostly from Germany, Poland and China, with minor collections from other (mostly European) countries least incomplete survey: http://php.indiana.edu/~donbyrd/MusicTestCollections.HTML what happens to music corpora after a while sad case of Princeton Josquin project (1970s) most punchcards are lost what remains is unusable

What if you want to work with encoded notation: 

What if you want to work with encoded notation developing yet another encoding system may not be very wise most encoding systems are richer than necessary for most music IR applications get scores (as ‘ground truth’ for signal processing?) collect from the Internet conversion not too many quality issue encode scores extremely timeconsuming OCR gets better and better work of Ichiro Fujinaga (will be at ISMIR) still relatively high error percentages