Voice conversion results using mapped MFCC Yprim (from Yannis).
The Yprim are aligned by DTW on the real Y MFCC (the original target).
Frame selection by Viterbi algorithm that doesn't allow Contiguous frame ! (Thierry's trick)
The excitation of the target is used to synthesize (LPC synthesis) the speech with the AR vectors selected


SUmmary:

- Use Yprim
- Yprim aligned by DTW on Y
- Use Target excitation
- frame selection by Viterbi and not allowing contiguous frames
