Joint lexicon, acoustic unit inventory and model design

Although most parameters in a speech recognition system are estimated from data by the use of an objective function, the unit inventory and lexicon ar e generally hand crafted and therefore unlikely to be optimal. This paper p roposes a joint solution to the related problems of learning a unit invento ry and corresponding lexicon from data. On a speaker-independent read speec h task with a 1k vocabulary, the proposed algorithm outperforms phone-based systems at both high and low complexities. (C) 1999 Elsevier Science B.V. All rights reserved.