公开(公告)号：DE3681023D1

公开(公告)日：1991-09-26

申请号：DE3681023

申请日：1986-03-27

Applicant: IBM

Inventor： BAHL LALIT RAI ,  DESOUZA PETER VINCENT ,  MERCER ROBERT LEROY ,  PICHNEY MICHAEL ALAN

IPC: G10L15/14 ,  G10L5/06

Abstract: The speech recognition system has a circuit for generating an alphabet of standard labels in response to a first speech input. Each standard label represents a sound type corresponding to a given interval of time. A circuit produces a respective sequence of standard labels from the alphabet in response to the uttering of each word from a vocabulary of words. A circuit selects a set of personalized label representing a sound type corresponding to an interval of time. A circuit forms a respective probalistic model for each standard label. Each model includes a number of states and at least one transition extending from a state to a state. It also includes a transition probability for each transition and, for at least one transition, a number of output probabilities. Each output probability at a given transition in the model of a given standard label represents the likelihood of a respective personalized label being poduced at the given transition.