A noise-robust speech recognition method composed of weak noise suppression and weak Vector Taylor Series Adaptation
2009 ◽