ByNobleID
    Incorporating a mixed excitation model and postfilter into HMM-based text-to-speech synthesis | NobleID