Spatial Organisation of Music

 

Kristoffer Jensen and Kirstin Lyon

 

Interviews with professionals and music lovers indicate that manual 

organisation is the only option if they need to re-find a piece of 

music efficiently. This is possible in the optics (?  maybe a 

different word? ) of a slow increase of the music collection, by 

buying records one at a time. The modern p2p sharing permits a much 

faster increase of the music collection. In addition, a shift from 

property payment to playback payment is easy to envisage in the 

future. This would allow anybody to have access to the total music 

production made to date, and only pay if a song is downloaded. A very 

large music database certainly demands help in organisation, browsing 

and retrieval if all music should be the potential next piece to 

listen to.

 

A simple organisation scheme is a two-dimensional spatial 

organisation where each new song can be placed in an arbitrary 

position on the workspace. The organisation of music will help in 

reducing the time and effort spent in re-finding individual music, as 

well as improving the familiarty of the database. This individual 

space organisation will be enhanced in the following ways; 1) 

clusters of organisations are identified with an automatic clustering 

algorithm, and landmarks are positioned in the centre of each 

cluster. An option to provide a verbal attribute to each cluster is 

given. 2) In order to facilitate  collaborative sharing and 

discussions of the music collections, each individual spatial 

organisation is scaled and rotated to minimise the distance to 

another organisation. This enabled the constructuion of one reference 

organisation, that has the minimal distance to all individual 

organisations. 3) includions of new music by inserting at the minimal 

distance to similar music. This is propsed by calculating the 

relationship of the spatial organisation ot meta-data and acoustic 

features, that would allow the simple insertion at the minimum 

distance to the features, mapped to the individual organisation.

 

At the moment, some experiments have completed to show the validity 

of spatial organisation. More work remains before steps 1) to 3) can 

be evaluated formally.