Speaker adaptation for telephony data using speaker clustering

This paper reports an ongoing effort to develop an unsupervised an-line speaker adaptation method for telephony environment. All speakers in the training data corpus are acoustically pre-clustered into clusters, and a cluster-dependent system is built for each duster. When a new telephony test speaker is given, a cluster, which is the closest to the speaker, is determined and selected by an improved distance measure. Based on this selected cluster, MLLR adaptation algorithm with block diagonal transformation is applied to move the cluster model to be closer to the testing speaker. For telephony application the adaptation data can be very short or noisy, potentially, the MLLR adapted means can be unreliable. A MAP-like weighting scheme for MLLR adaptation is applied to insure the adapted mean reliable when the adaptation data is very short.