Beat Explorer Content Based Search for MIDI File Format

This project aims on a variety of music search methods for users. It allows users from a
variety of backgrounds and capabilities to use the system. It provides searching capability
using humming, music note and parson code for midi file format. Developers and users
require different method in querying and retrieving database information with a rich variety
of methods to enhance the capability of the system. For the advance users, especially
composers, searching by music note and parson code would be the best choice for them since
it would be able to provide more precise output while for other users, humming is the best
choice since almost everyone can hum. Since not everyone could remember the song title,
humming provide the searching capability based on the melody they hum. The system will
record the humming sound and filter the noise before process it to produce the frequency of
· humming. The frequency value will be converted into music contour in parson code to be
compared to the data in the database. For music note, it will also be converted into frequency
and music contour for comparison. The system will output the song title together with the
similarity ratio of the music contour. This project uses waterfall model approach for the
development due to lots of advantages with the appropriate development tools. It is
developed using a JAVA technology and open source frameworks together with library
provided from SUN Microsystems.