ABSTRACT
This chapter summarizes what are considered to be the nine most important research topics, in particular those which intertwine a multiplicity of speech research areas. For the majority of humankind, speech understanding and production are involuntary processes quickly and effectively performed throughout their daily lives. A part of these human processes has been synthetically reproduced, owing to the progress in speech signal processing, linguistic processing, computers, and LSI technologies. Although individual aspects of speech information processing research have thus far been performed independently for the most part, they will encounter increased interaction until commonly shared problems become simultaneously investigated and solved. The voice individuality problem in speech recognition has been handled to a certain extent through studies into automatic adaptation algorithms using a small number of training utterances and unsupervised adaptation algorithms. Speaker adaptation or normalization algorithms in speech recognition as well as speaker recognition algorithms should be investigated using a common approach.
