作者: Makoto Shozakai , Goshu Nagino
DOI:
关键词:
摘要: To provide a data process unit and control program that are suitable for generating acoustic models unspecified speakers taking distribution of diversifying feature parameters into consideration under such specific conditions as the type speaker, speech lexicons, styles, environment providing intended adapted to person. The comprises classification section, storing pattern model mathematical distance calculating converting display region dividing division changing selecting section.