A new technique of speaker adaptation for use in speaker-independent speech recognition systems is presented. The training-data is used to build models (based on linear regression) of sounds. At recognition time, the models are used together with an incomplete set of sounds from a new speaker to estimate values for unheard sounds, which are then used to adapt the speaker-independent models. The technique reduced the error-rate from 17% to 5.3% when applied to a database of 104 speakers speaking the English alphabet.
Bibliographic reference. Cox, Stephen (1993): "Speaker adaptation using a predictive model", In EUROSPEECH'93, 2283-2286.