PSOLA like prosody modification algorithms for harmonics plus noise models.
The major difference is that in HNM, we repeat or eliminate speech frame parameters to realise prosody modifications
instead of directly using speech waveform itself.
The synthesis stage performs appropriate interpolations among successive frames to ensure continuity.