Speech recognition: Current status and prospects
Li Deng · Canadian acoustics · 1998
A critical review of the state-of-the-art automatic speech recognition is provided which include: analyzing the fundamental equation of statistical speech recognition; fundamental architecture of the modern speech recognition systems dissected into their common, basic phonetic and phonological components; and critical review of the backbone of modern speech recognition technology, hidden Markov method (HMM). A number of key research issues which are aiming to overcome several fundamental deficiencies in the current speech recognition technology are addressed. An approach is outlined in which a version of nonlinear phonology are constructed from multidimensional features overlapping in time.