A new band-splitting method for two-band speech model

摘要

A new band-splitting method for two-band speech model is proposed,nwhereby a score function based on the subband periodicity is calculatednfor each harmonic band and the band of maximum score is selected as thenband-splitting frequency. Using the score function, a tracking techniquenis applied to the algorithm for the continuity of the band-splittingnfrequencies between neighboring frames. This method has proven to benrobust to the inaccuracy of pitch estimation and the frequencynresolution problem in low-pitched speech