Large-vocabulary recognition

Large-vocabulary continuous-speech recognition (CSR) technology is at work. As an application of the technology, we will describe a dictation system (DS). Input to the system is unrestricted spontaneous speech. No adaptation, no special skills are required to use the system. The DS transforms continuous speech into written text. It is essential in this application that the user is free to speak as he or she usually does and should be free to use his or her own wording and formulation. This implies speech recognition for large and open vocabularies, free syntax, continuous speech. The aim of the paper is an attempt to determine what is feasible with today's technology and what will be feasible in the near future. The problems addressed are: what are the limits of today's technology, what is needed to make the next step, i.e. going towards real industrialization of CSR technology.