AN IMPROVED GLOBAL SOFT DECISION METHOD INCORPORATING SECOND-ORDER CONDITIONAL MAP FOR SPEECH ENHANCEMENT

摘要

PURPOSE: A second CMAP base broad area soft decision algorithm for a speech enhancement method for derivation of the voice based on 2 CMAP is provided to presume the noise in noise environment which changes the time based on the second CMAP considering the voice activity of the adjacent frames. CONSTITUTION: A CMAP(Conditional Maximum A Posteriori) about the voice presence and member defines in consideration of the interconnectivity of the adjacent frame(S100). The speech absence probability of the current frame obtains based on CMAP about the voice presence defined as a member(S200). The voice of the current frame applies the speech absence probability(S300).