High-Level Music Content Description
An intuitive starting point for content-based music information retrieval is to use musical concepts such as melody or harmony to describe the content of the music. In the early days of MIR, many query-by-humming systems were
proposed that sought to extract melodic content from polyphonic audio signals (those with multiple simultaneous musical lines) so that a user could search for music by singing or humming part of the melody; such systems are now being deployed as commercial services; see, for example, naiyo.com.
A survey of sung-query methods was conducted by Hu and Dannenberg in.
High-level intuitive information about music embodies the types of knowledge that a sophisticated listener would have about a piece of music, whether or not they know they
have that knowledge:
BIt is melody that enables us to distinguish one work from another. It is melody that human beings are innately able to reproduce by singing, humming, and whistling. It is melody that makes music memorable: we are likely to recall a tune long after we have forgotten its text. Even though it is an intuitive approach, melody extraction from polyphonic recordings, i.e., multiple instruments playing different lines simultaneously, remains extremely difficult to achieve.
Surprisingly, it is not only difficult to extract melody from audio but also from symbolic representations such as MIDI files. The same is true of many other high-level music concepts such as rhythm, timbre, and harmony. Therefore, extraction of high-level music content descriptions is a subgoal of MIR and the subject of intensive research.
The goal of such tasks is to encode music into a schema that conforms to traditional Western music concepts that can then be used to make queries and search music.
