Pronunciation variants description using recognition error modeling with phonetic derivation hypotheses

This paper proposes a new method of pronunciation variant generation for reducing word error rate in conversational speech recognition. In particular, this paper focuses on the generation of alternative pronunciations from canonical forms by using the phonological knowledge derived from the analysis of a phonetic transcription corpus. The experimental results show that the pronunciation variation generated by the proposed method provides slightly better performance than a method based on manually written pronunciation. These results also demonstrate the applicability of phonological knowledge-based generation of pronunciation variation.