I wish to have information as to the search paradigm of the new melody search, the repertoire included, is it derived from notes or recordings, and how it generally works
Yossi Goldenberg
head librarian
Jerusalem Academy of Music and Dance
IMSLP melody search?
Moderator: kcleung
-
pml
- Copyright Reviewer
- Posts: 1219
- Joined: Fri Mar 16, 2007 3:42 am
- notabot: 42
- notabot2: Human
- Location: Melbourne, Australia
- Contact:
Re: IMSLP melody search?
Hopefully the designer of the search system will see and answer your post. An article on the IMSLP journal is expected soon.
It looks like optical music recognition software has been applied to most of the IMSLP corpus of scores (excluding technical blocks on works in copyright both in the US and EU, or works not in standard notation) to generate the data set, which consists of sets from 2 up to 11 or so notes organised by semitonal jumps. Starting pitches tends to be preserved (including transposition in the case of instrumental parts for transposing instruments). The generalized search appears to be key insensitive. Grace notes or embellishments of melodies don’t seem to have been recorded and included in sequences, but one would have to test this!
As for works with multiple lines sharing staves, or the appearance of vertical homophobic chords, it's a guess whether all lines are sequenced or just the highest sounding pitch.
The search is obviously not recording-based.
Cheers, Philip
It looks like optical music recognition software has been applied to most of the IMSLP corpus of scores (excluding technical blocks on works in copyright both in the US and EU, or works not in standard notation) to generate the data set, which consists of sets from 2 up to 11 or so notes organised by semitonal jumps. Starting pitches tends to be preserved (including transposition in the case of instrumental parts for transposing instruments). The generalized search appears to be key insensitive. Grace notes or embellishments of melodies don’t seem to have been recorded and included in sequences, but one would have to test this!
As for works with multiple lines sharing staves, or the appearance of vertical homophobic chords, it's a guess whether all lines are sequenced or just the highest sounding pitch.
The search is obviously not recording-based.
Cheers, Philip