A speech recognition method and apparatus is disclosed in which each utterance matched to a feature model within a feature model memory (110) is associated with a confidence score indicative of the posterior probability of the word being correctly matched, given that the matching of the utterance to the feature model generated certain values indicative of the goodness of the match. The confidence score for the matching of an utterance to a feature model is determined from the generated values indicative of the goodness of the match and a stored set of parameters indicating the probability of the generated values arising given that a match is either correct or incorrect.
A control apparatus controls the display of text data which includes a speech link that can be activated by spoken command. The shape of a pointing device cursor displayed on a display is then changed by the apparatus when the pointing device cursor is located over the speech link included in displayed text data. The apparatus is arranged to output a prompt identifying speech commands that can be used to activate the speech link if the pointing device cursor is displayed on a display located over the speech link in a changed state for a predetermined time.
