METHOD FOR SPEAKER ADAPTATION BASED ON MAXIMUM A POSTERIORI EIGENSPACE

摘要

The present invention relates to a speaker adaptation method based on a maximum post-specific space, if due to the speech recognition of differences in learning the speaker and recognized the speaker adaptation data, a small amount given to overcome the degradation of the speech recognition adaptation based on the speaker space improve techniques learning screen or by using a line speaker spatial information obtained from a particular speaker by using adaptation binding to the new adaptation data can improve the recognition performance for mismatch environment between learning the speaker and recognized speaker and the specific speaker recognition system this has the advantage that improve adaptation to.