Content Based Music Retrieval

Two main groups of Music Information Retrieval (MIR) systems for content-based searching can be distinguished, systems for searching audio data and systems for searching notated music. There are also hybrid systems that first transcribe audio signal into a symbolic description of notes and then search a database of notated music. An example of such music transcription is the work of Klapuri [10], which in particular is concerned with multiple fundamental frequency estimation, and musical metre estimation, which has to do with ordering the rhythmic aspects of music. Part of the work is based on known properties of the human auditory system. Content-based music search systems can be useful for a variety of purposes and audiences: • In record stores, customers may only know a tune from a record they would like to buy, but not the title of the work, composer, or performers. Salespeople with a vast knowledge of music who are willing and able to identify tunes hummed by customers are scarce, and it could be interesting to have a computer do the task of identifying melodies and suggesting records. • A search engine that finds musical scores (notations of musical works) similar to a given query can help musicologists find out how composers influenced one another or how their works are related to earlier works of their own or by other composers. This task has been done manually by musicologists over the past centuries. If computers could perform this task reasonably well, more interesting insights could be gained faster and with less effort. • Copyright infringement could be resolved, avoided or raised more easily if composers could find out if someone is plagiarizing them or if a new work exposes them to the risk of being accused of plagiarism. A content-based music retrieval could facilitate such searches. Content-based search mechanisms that work specifically for audio recordings can be useful for the following purposes: • It is possible to identify music played, for example, on the radio or in a bar by pointing a cellular phone at the speakers for a few seconds and using an audio fingerprinting system for identifying the exact recording that is being played. • Recordings made by surveillance equipment can be searched for suspicious sounds. • Content-based video retrieval can be made more powerful by analyzing audio content, including music. • Theatres, film makers, and radio or television stations might find a search engine useful that can find sound effects similar to a given query or according to a given description in a vast library of audio recordings.