Fields of automatic speech recognition

Speech recognition is an almost contemporary discipline in computer science. Around 1950, the first digit recognition system appeared, an entirely wired and highly imperfect device. Around 1960, the introduction of digital methods and the use of computers changed the dimension of research. Nevertheless, the results remained modest, as the difficulty of the problem had been greatly underestimated, particularly where continuous speech was concerned. Around 1970, the need to use linguistic constraints in the automatic decoding of sentences became clear, whereas speech recognition had hitherto been considered an engineering problem. The end of the 1970s saw the completion of the first generation of commercialized word recognition systems. Subsequent generations, taking advantage...