An autoencoder neural-network based low-dimensionality approach to excitation modeling for HMM-based text-to-speechSrikanth VishnubhotlaRaul Fernandezet al.2010ICASSP 2010
