Have you ever found yourself unconsciously tapping your foot while
listening to music? Ever wonder how it is that humans can intuitively
detect the beat of a song? This project analyzes the beats of two
different musical excerpts and speeds up the tempo of the slower
signal to match that of the faster one. Traditional manipulation
of signal duration would normally cause alteration of the signal's
frequency, most commonly evidenced in the way a musical signal sounds
higher or lower when played faster or slower. This effect is obviously
less than ideal if the original musical quality of the excerpt needs
to remain the same. The goal of this project is to determine the
beats of two songs and then utilize frequency analysis techniques
to change the duration of those two songs such that they were equal
without causing alterations in the pitches of either
Download our PowerPoint Presentation.
Such a beat synchronizer would be of value in many applications
such as human-computer improvisation, music transcription, music
editing, and ethno-musicological studies. It would open up possibilities
for computer in music by allowing the computer to synchronize with
tempos external to it without the use of explicit synchronization
information. Musicians generally agree that it is more difficult
to play in an ensemble and follow a collective tempo than it is
to set their own tempo and require that other musicians follow.
Similarly, when disc jockeys remix and edit different songs, such
a synchronization system becomes a necessity if aurally-pleasing
sounds are to be generated for the listener. In such a situation,
a beat synchronizer would take the musical input of two different
songs, analyze their beat frequencies, and produce two songs whose
beats were synched.
A beat synchronizer could also be used to determine rhythmic similarities
in music. Rhythmically similar music will have similar beat spectra.
A measure of similarity could be computed by comparing the beat
spectra. Among other applications, this allows retrieval by rhythmic
similarity: a collection of music could be ranked by similarity
to a given musical example. Another application might be to arrange
songs by similar tempo and rhythm so that the transition between
them is smooth. Studies are also performed to analyze the different
rhythmic characteristics displayed in the traditional songs of various
ethnicities. Such information can be useful in learning the evolution
of music with culture over a given period of time.
Additionally, automatic tempo extraction can be accomplished through
a beat synchronization system. Knowing the song structure, tempo,
and beat times allows synchronization of external events to the
audio. For example, an animated character could nod or dance to
the musical tempo. Video clips could be automatically sequenced
to match the tempo of an chosen musical soundtrack.