Voice conversion results using mapped MFCC Yprim (from Yannis). THe frame selection is done by Viterbi algorithm (preferring contiguous frames), and not allowing the contiguous frames. (The trick of Thierry). 
The excitation of the source is used to synthesize (LPC synthesis) the speech. 


Summary:

- Use Yprim
- Use Source excitation
- frame selection by Viterbi (not allowing contiguous frames) 



