What is meant by “continuous” and “discrete” speech recognition?
A. Continuous speech systems enable a competent user to dictate without pausing between words and phrases. The earlier discrete systems required a small pause of about a tenth of a second between words, resulting in a slightly stilted delivery. Speeds of well over 120 words a minute of accurately recognized text can be achieved. Dictation must be well articulated but not artificially so. Any tendency to mumble, gabble or slow down the speech will reduce the accuracy of recognition considerably. In general, users of speech recognition will not generate text at such high speeds. Professional typing speeds rarely exceed 80 words a minute. Most people who use dictation equipment achieve speeds of around 100 words a minute. Therefore. For practical purposes DNS will cater for all speaking speeds.
