Main Menu

Research & Integration


WP2: Evaluation, Integration and Standards
WP3: Visual Content Indexing
WP4: Content Description for Audio, Speech and Text
WP5: Multimodal Processing and Interaction
WP6: Machine Learning and Computation Applied to Multimedia
WP7: Dissemination towards Industry

Logo IST
Map of Mozart

Mozart's complete works analyzed

Andi Rauber's team focuses on research for Music Information Retrieval and thus on methods which enable automatic recognition of style, genre or artist of a piece of music or retrieval of similar music. Moreover, the information extracted by these methods builds the basis for a novel way of access to music archives.



Map of Mozart
Map of Mozart


For the Map of Mozart project they analyzed the complete works of Wolfgang Amadeus Mozart based on the audio content (not on the musical notation!). The computer is able to detect the main characteristics of each piece and attempts to organize the works by musical similarity. Rauber's team uses a Self-Organizing Map (SOM), an unsupervised learning algorithm, for organization of the pieces of music. The result is a map with different regions showing distinct categories of Mozart's music, while similar music is grouped close to each other. The map's shape has been chosen to resemble the shape of Mozart's head. The outcome is - a Map of Mozart.

The map arranges the operas written by Mozart in one specific area, while placing piano pieces in an opposite area. String ensembles, violin sonatas, serenedes - each region of the Map of Mozart contains a group of works with specific characteristics, with smooth transitions between the groups. This visualization provides an immediate and intuitive overview of the complete works of Mozart. Furthermore, it enables interactive browsing of the Mozart archive and retrieval of additional information about the works of the world-famous artist.

This website contains an Online-Demo of the Map of Mozart. The web interface currently only features limited functionality: It provides different visualizations and the possibility for selecting areas of the map in order to retrieve information about the titles located there. Due to license issues we cannot provide the complete collection of the actual music which has been mapped, nevertheless we it is possible to hear some pieces available from public domain or under the Open Audio License (from Wikipedia and Internet Archive).

The PlaySOM application we developed provides extended functionality and enables a completely novel way of enjoying music: Playlists can be created by navigating and drawing paths on the map. This sort of interface allows the creation of personal playlists for by directly selecting regions of one's personal taste, without having to browse a list of available titles and manually sorting them into playlists.

Additionally we developed the PocketSOMPlayer, which brings the Maps of Music to portable devices (PDAs, portable audio players, mobile phones). The quick creation of playlists to the user's taste definitely improves the experience of portable music.