We assume that, with some practice, Mark would be able to modify the
constrastive stress into ordinary nuclear, and that he'd be willing to put
in the practice necessary, so we don;t see that as a major problem.
We think we could live with the automatic formant tracking problem, but
it's far from ideal. The tracker is normally good, and we've checked
some of the same phrases with Ali, Sarah, and Jonny Rodgers' speech, and
it does better on all of them. (I can hear Mark saying "I did warn you"!
And yes, I did say I had me doubts too, I know.) In most (but not all
cases) hand-done spectra compensate.
But the other things are a real problem. They include: much breathy voice,
much creaky breathy stuff at the ends of vowels, marked shimmer, and
probably marked jitter phrase finally (amplitude
modulation [which may throw the tracker off - unsure yet], and f0
modulation). Not to mention a relatively unclear formants.
Suggestions:
1. Jonny Rodgers is trying his hand (mouth?) at it. But he devoices all
his vowels and has a breathy voice too, so he's practising a bit.
2. Mark: is there any chance you could develop a better spectrogram
voice? Ali and I have both done that: being but women we felt we had to.
We feel that if you could get rid of the constrastive stress, and try to
use a more modal voice throughout, you might be a winner. Perhpas we could
have a go (with acoustic analysis) when we get together?
Got to rush now, but will send notes on specific files tomorrow if you
feel it's worth it.
JohN: when you said you thought the voice was exactly what we needed, had
you done any acoustic analyses? Either way, could you comment please?
Sarah
______________________________________________________________________
Dr. Sarah Hawkins Email: sh110@cam.ac.uk
Dept. of Linguistics Phone: +44 1223 33 50 52
University of Cambridge Fax: +44 1223 33 50 53
Sidgwick Avenue or +44 1223 33 50 62
Cambridge CB3 9DA
United Kingdom