Smoothed local adaptation of connectionist systems

abbot is the hybrid connectionist hidden Markov model (HMM) large vocabulary continuous speech recognition system developed at Cambridge University Engineering Department. abbot makes e ective use of the linear input network (LIN) adaptation technique to achieve speaker and channel adaptation. Although the LIN is e ective at adapting to new speakers or a new environment (e.g. a di erent microphone), the transform is global over the input space. In this paper we describe a technique by which the transform may be made locally linear over di erent regions of the input space. The local linear transforms are combined by an additional network using a non-linear transform. This scheme falls naturally into the mixtures of experts framework.