Well, no "explanation" on web page below, just file names...
>The best 2 are very similar - just one small mystery change for you
>to identify. If you notice and like the change, we can perhpas try other
>small ones that might also help.
Best 2 *are* impressive, though my listening sensitivity is impaired by
builders on roof above. Would welcome further clues as to where I should
direct listening and will try to rig up headphones. Currently I have a
marginal preference for hand2 over hand1, but hand1 is also good so it's a
case of which adjustments generalise better. (hand1 seems to have a tiny
break in the f0 during the diphthong, but under current conditions I may
just be hearing things).
>http://kiri.ling.cam.ac.uk/heid/cosy.html
>
>This work forms the centre of the Heid and Hawkins paper for the ESCA
>synthesis workshop (paper due in Ipswich 12 September). The copy synthesis
>is not perfect, and it's certainly not a perfect reproduction of Mark's
>voice quality, but we think the intelligibility is good, and the
>naturalness is not too bad.
It's good to have something to listen to! Look forward to discovering how
it relates to our structure...
>
>We would be grateful for your comments about the quality of the best
>one(s) we've done, and your ideas for further improvements, before we
>(more precisely, Sebastian!) writes further rules to include our
>hand-adjustments, and we see if the rules generalize to other utterances.
>(After that, we want to break down some of the natural coarticulation.) If
>you could reply ASAP, due to imminent due date for paper above, it would
>be very helpful.
>
Well done. I don't have specific ideas for further improvements at this
point, but look forward to seeing/hearing how "rules" generalise to other
utterances.
Jill