Katsuteru MARUYAMA, Masami NAKAMURA,Takeshi KAWABATA and Kiyohiro SHIKANO
English Word Recognition Using HMM Phone
Concatenated Training and NETgram
Abstract:For realizing an interpreting telephony system, an accurate word recognition
system is necessary. Because it is difficult to recognize English words using only
their acoustical characteristics, an accurate word recognition system needs
certain linguistic information. The NETgram, which is trained using Brown
Corpus is applied to HMM English word recognition. NETgram can correct word
recognition errors effectively. NETgram improved the word recognition rate from
81.0 to 86.9%. The performance is improved over that of traditional statistical
trigram models.