Design of a music retrieval system based on vocal queries

Currently users find their desired song on a multimedia sharing website such as YouTube using text based queries. The search phrase usually contains some text information about the song such as the name of the song, the singer, or parts of the lyrics. However, this information can be easily manipulated by malicious file up-loaders, who can use this scheme to upload pirated assets. This problem can be eliminated by letting users search based on the melody of the song rather than using textual information. The purpose of this project is to design a system that can help BroadbandTV, a Vancouver-based high-tech media company, with finding copies of music or music videos in an online video database such as YouTube. During this internship, I am planning to implement a query by humming (QBH) system, which achieves the above goal. Upon successful implementation, the system should be able to detect a song, whether it is an exact copy or it has been hummed by a user.

Faculty Supervisor:

Dr Rabab K. Ward

Student:

Mani Malek Esmaeili

Partner:

BroadbandTV Corp.

Discipline:

Engineering - computer / electrical

Sector:

Digital media

University:

University of British Columbia