TD-PSOLA versus harmonic plus noise model in diphone based speech synthesis | IEEE Conference Publication | IEEE Xplore