Voice source modelling using deep neural networks for statistical parametric speech synthesis | IEEE Conference Publication | IEEE Xplore