An ESOTerIC approach to some problems in automatic speech recognition

Abstract

An outline of the overall problem of automatic speech recognition-the ASR problem-has been given in previous papers (Hill, 1967a, 1968). The problem divides into three main areas: (1) Problems of definition; (2) Problems of implementation; (3) Problems of application. The present paper describes a machine, “ESOTerIC”, which embodies, in an elementary form, sections corresponding to those of the “basic acoustic recognizer” proposed in those papers. Some experimental results obtained with this machine are discussed, and the approach adopted is generalized. Finally, the possibility of immediate applications for this simple ASR system is discussed.