ODYSSEY 2004 - The Speaker and Language Recognition Workshop
May 31 - June 3, 2004
In previous work , it was investigated how the neighborhood can be used to estimate a better model for a speaker when few training data is avalaible. In this paper, this work is completed by investigating another way to merge models from the neighbors and by introducing a weight on the neighbor models to be merged. Experiments on a telephone speech database show that using the neighborhood-merged model to initialize the training phase provides improvement compared to the UBM approach, when few training data is available.
Bibliographic reference. Charlet, Delphine (2004): "Neighborhood-adapted GMM for speaker recognition", In ODYS-2004, 227-230.