Music information retrieval: creating an open source melody transcriber

Music information retrieval (MIR) is the fi eld dedicated to the organization and retrieval of music data. One of the eventual goals of MlR research is automatic transcription, where any musical recording can be transcribed to accurate sheet music with no human intervention, allowing easy cataloging of music and facilitating the creation of music databases for query-by-example systems. Towards this goal, l have investigated pitch detection algorithms for hummed or instrumental input and created an open source tool to transcribe monophonic melodies using the Java programming language. Additional goals include examining the current state of MlR research and encouraging future graduate students to explore MIR and expand on this tool.