ACOUSTIC SEPARATION METHOD AND APPARATUS

摘要

An acoustic separation method is disclosed. According to an embodiment of the present invention, the acoustic separation method comprises the following steps of: classifying random acoustic information by category to generate an acoustic signal; dividing each acoustic signal by category into a plurality of frames having a predetermined length; converting the plurality of frames into a frequency domain to generate a plurality of frame spectra and generating an acoustic dictionary made of a plurality of representative values about the generated plurality of frame spectra; and separating a sound by each category by using the acoustic dictionary if a mixed acoustic signal, in which sounds of the plurality of categories are mixed, is inputted.