Language discrimination device

摘要: PURPOSE: To improve the language discrimination capability by grasping differences in phoneme pronunciations.spectrum structures languages through use of an ergodic HMM for each and also array structure condition group which is common to all languages. CONSTITUTION: A feature extracting section 1 converts text audio signals into series vectors. An generating 5 generates from vector extracted using sounds various kinds as learning stored a storage 6. optimum calculating 7 obtains corresponding series. trigram 10 model generated sections 11 13 stored. COPYRIGHT: (C)1995,JPO