Two-step correction of speech recognition errors based on n-gram and long contextual information.

Ryohei Nakatani, Tetsuya Takiguchi, Yasuo Ariki

Browse the full Interspeech paper archive.