5th International Conference on Spoken Language Processing
Although most parameters in a speech recognition system are estimated from data, the unit inventory and lexicon are generally hand crafted and therefore unlikely to be optimal. This paper describes a joint solution to the problems of learning a unit inventory and corresponding lexicon from data. The methodology, which requires multiple training tokens per word, is then extended to handle infrequently observed words using a hybrid system that combines automatically-derived units with phone-based units. The hybrid system outperforms a phone-based system in first-pass decoding experiments on a large vocabulary conversational speech recognition task.
Bibliographic reference. Bacchiani, Michiel / Ostendorf, Mari (1998): "Using automatically-derived acoustic sub-word units in large vocabulary speech recognition", In ICSLP-1998, paper 0586.