Learning continuous representation of text for phone duration modeling in statistical parametric speech synthesis | IEEE Conference Publication | IEEE Xplore