Duration Distribution Based MFM Model for Speech Recognition | IEEE Conference Publication | IEEE Xplore