Real-time perceptual tempo estimation for music signal based on envelope autocorrelation

Tempo and beats are very important elements in the perceptual music. This paper proposes a real-time perceptual tempo estimation algorithm, which calculates the envelope autocorrelation of the music signals firstly, and then a weighted tempo function based on envelope autocorrelation is constructed to estimate the music's perceptual tempo. The weighted tempo function is computed according to the peak amplitudes and positions of the envelope autocorrelation, which can demonstrate the real-time tempo information of the music signal, and its statistics can be used to represent the rhythm of the whole piece of music. Our experiment results show that the proposed weighted tempo function is highly related to the sense of auditory rhythm, and is capable of describing the music tempo changes at real-time.