Chord Ontology:Development
From Music Ontology Wiki
Contents |
[edit] Introduction
The current Chord Ontology ('draft 1') is described at http://purl.org/ontology/chord/ (or http://purl.org/ontology/chord/draft1/ in perpetuity).
This page is for summary and discussion of proposed changes or additions for draft 2 of the ontology.
[edit] Proposed changes
[edit] Changes to the chord description model
- Fix : Allow unnamed root note - Currently intervals may be pitch class rather than named note, by using a SemitoneInterval rather than a ScaleInterval, but the same cannot be done for the root of the chord.
- How to specify pitch of root ? Number 0-11 with 0=C ? see also point below about attaching notes to intervals.
- Add : Allow root to be specified as a degree of the key
- "One thing I think might be missing is to allow scale degree notation as is very common in musicologists' work, i.e. they describe C Em F Ab as I iii IV bVI" -- MatthiasMauch
- Add : Explicit information about where chord information has come from.
- This should be possible to infer from context
- but explicit labelling like "human annotation", "machine annotation", "musical score" might be helpful in some situations ?
- Specifying a confidence value in a chord:ChordEvent or chord:Chord instance might also be helpful
- Add : Specification of chord voicing
- Common in jazz to specify eg. open vs. closed. vs. mixed voicing
- Add : Associate intervals of a chord with actual note resources
- Possible tie-in with the symbolic notation ontology ? <http://motools.svn.sourceforge.net/viewvc/motools/so/>
- This could help with the 'unnamed root notes' issue above
- This also removes the need for the chord ontology to cover specification of component notes by octave and pitch class (as had been suggested)
- Add : Higher level information about the functions of chords
- Incorporating work from our colleagues Tillman Weyde and Jens Wissmann at City should take care of this nicely
- Issue : Metrical time
- Hopefully this will be developed further in either the TimeLine or symbolic notation ontologies.
- Specify : semantics of base_chord relations, and the 'shorthand' symbols (eg. using <http://purl.org/ontology/chord/symbol/C:maj7> rather than <http://purl.org/ontology/chord/symbol/C:(1,3,5,7)>)
- Depending on who you ask, these are either :
- Just a shorthand so you can type C:maj7 instead of C:(1,3,5,7)
- An additional semantic label, specifying the interpretation of the coincident intervals (1,3,5,7) as a major seventh chord.
Personally I feel only the second option has any place in an RDFS ontology, and I think we need these semantically-weighted labels in the ontology. Taking this interpretation means that some of our data (eg. the Beatles transcriptions) must be re-converted because the original use of the shorthands in Chris Harte's notation does *not* necessarily imply any interpretation on the intervals.
--Chris Sutton 15:58, 3 December 2007 (CST)
- Reconsider the name ScaleInterval
- Perhaps misleading - eg. is the ScaleInterval with degree 3 in a major chord different from the ScaleInterval with degree 3 in a minor chord ? (answer: no, but the name suggests in might be)
[edit] Extra information in symbol service descriptions
- Add : Guitar tab information to chord descriptions
- Add : Score snippet images to chord descriptions
- Done and live now, courtesy of [Christophe Rhodes] at Goldsmiths
- eg. <http://sonictruths.net/junk/chord.png>
- Reconsider : chord symbol service notation
- This was originally taken directly from Chris Harte's notation, switching the '#' for a 's'
- Would be better to avoid using '/' for bass notes
- Might be better to avoid commas
[edit] Documentation
- Add : Use case descriptions
- We have a few use cases which could be written up
- At QM : conversion from QM notation .lab files, mma files, or Goldsmiths notation CSV files for consumption by Java chord analysis suite
- John Ibbotson's work on semantic analysis of MIDI streams
- any others so far ?
- Danny suggested :
- transcribe chordal music ?
- play it back ?
- Similar to use cases, some 'classic' chord examples might be useful, to show how you'd describe particular chords using the chord ontology.
- eg. the ['Tristan chord'] from Tristan & Isolde or some quartile harmony examples
- We have a few use cases which could be written up
- Add : tests
- How to design these ?
- Add : example data sets
- mma set should be published
[edit] Other ontologies
See also : Key_Ontology:Development for discussion of describing scales and keys.

