Joe: Music Notation CG update
... will talk about the CG, why it exists, what the current projects are
... create a next generation music notation markup
... will look at some of the fundamental problems in music notation
... I'm the CEO of noteflight.com, a division of one of the largest sheet music publishers
... we have a web based editing and playback environment
(demo'd a web based online score with a "play" button that uses webaudio to play the score)
... we're in a place now that music notation is being used all over the web and yet the approach to representing it is a 15+ year old approach
... the job of the CG is to figure out what to do to account for the new use cases now that music notation is on the web
... co-chairs of the CG are Joe, Michael Good, Daniel Spreadbury
... over 200 people have joined the group, so there's a lot of interest, although the group is not super active
... we want to get to a place where it's easier to develop notation applications
... the CG hasn't totally gotten off the ground yet
.. MusicXML represents the actual musical content
liam [ http://www.smufl.org/ ]
... SMuFL is the font system
... MNX will hopefully be the successor to MusicXML
... we want to make it mesh with things like CSS and the DOM API
... or the social APIs that the socialwg is working on
... MNX is kind of a "reboot" of MusicXML. the initial focus will be common western music notation
... people asked whether this will be a western centric standard or can it incorporate other notation systems
... it's very difficult to create a single unified framework to take all of these into account, so we're creating essentially a shell format that you can put things into, but then focusing on western notation
... let's go through a few examples of why music is a lot weirder to represent than you might think
... at its core, music notation is a series of instructions to a group of performers over a period of time
... most of the time you can get away with mostly just pitch/duration things, but let me show you a few counter examples
... this melody at the top begins with a sequence of triplets. in the very first bar, the "3"s under each grouping indicates the subdivision into triplets. then it continues, and the 3s get left out.
... normally when the 3s are left out this has a very different meaning. but the musician reading this understands that the 3s are just left out. this is a simple example of how the visual and semantic drift apart, and there's often an element of relying on the interpretation of the performer
... so you might imagine doing this with "display:none" on the later "3s"
... if you were doing this in finale or sibelius you probably would notate the "3s" and then hide them
... here's another example. here is part of a chopin nocturne, a very complex 2-part piano work. in the right hand there are two to three voices going on simultaneously. chopin has used beams and stem direction to show the separate strands of melody going on at the same time even though they are being played with one hand.
... the performer is meant to play these as different melodies with the same hand
... that note i highlighted in red, the last note in the bar, that note is meant to be a part of both melodies
... it looks like it's written as two overlapping notes at the same time, but it doesn't have a well defined meaning and is up to the performer to untangle this
... this one is much harder, you can't just achieve this by hiding stuff
... someone said "there's no way music notation can be represented by a tree structure, so this should always just be SVG and then use MIDI"
... you'd know how to play it because of the midi but you wouldn't know anything else about it
... other use cases: render notation as graphics or audio. it might be someone who can only listen to it. we might want to render it in musical braille.
... the needs of music on the web start to include much more than just showing it. you want to convey its content to the user in a bunch of different ways.
... "Happy Together" here is a view a conductor might see, so a conductor can direct a choir. here's a completely different view of the document. this is just one part that a single singer would see, they wouldn't want to see the piano part. this is a different view of the same music.
... just for fun, i rendered this in a different font
... the needs of a choir singer and a conductor are different. in the print world, these are created as separate documents.
... i'm pretty sure in our ideal world we would like this to be in one document
... that's the end of my presentation. we haven't gotten very far with MNX other than defining stories.
... i'd like to talk here about some of the problems we'd like to solve
jessebeach q+
jessebeach +q
... can we take advantage of style-like concepts of CSS, that will help us separate semantic notions from visual
jessebeach q?
* Zakim sees no one on the speaker queue
jessebeach q+
* Zakim sees jessebeach on the speaker queue
... if we make music notation a DOM, then what do we get from that and what kinds of problems do we run into especially from the non-tree aspects of music
... as the examples we've seen show, notation is partly an art, and composers are constantly breaking the boundaries, so it can be very difficult to say "what is semantics" of the music and what's styling
... there are some things that are pretty straightforward. if i see a quarter note, it feels like semantic musical content, you can find a way to render that either as audio or visual. there are some things that are very clearly style, like the font that is used for the noteheads.
... what about things like tweaking the heights so that things line up, is that semantic or styling
liam: some of that is semantic, showing that two parts of the music line up
joe: (looking at the chopin example) a lot of what is here is semantic.
joe: right now there isn't really anything between MIDI and raw audio
pauljessup: maybe the problem is trying to find a single representation for both an instruction to a performer and an instruction to a machine
