Cross-variety speaker transformation in HSMM-based speech synthesis




Samples for transformations between Austrian Standard (AT), Bad Goisern dialect (GOI) and Innervillgraten dialect (IVG).

 RecordedSpeaker dependent (SD)Data Mapping (DM)Constrained Phone Data Mapping (CPDM)Notable difference
AT to GOI No crackling in the "s" of "übers" (second last word).
AT to IVG "Nichts" is smoother, no crackling in middle of word.
IVG to GOI "Feld" (last word) is not corrupted.
GOI to IVG "gor et" smoother.