Duration modeling in automatic recited speech recognition | IEEE Conference Publication | IEEE Xplore