Discrimination between speech and music using time series events

In this paper, a novel method for the discrimination of speech and music, motivated by the zero crossing rate (ZCR) approach, is proposed. This method has the potential to detect the presence of music by a simple count of a certain event related to non zero crossing.