Re: Edinburgh presentation

Sarah Hawkins (sh110@cam.ac.uk)
Mon, 21 Sep 1998 15:45:48 +0100 (BST)

Response from Sarah, not mentioning hte things Jill has added - thanks
Jill. Sebastian is back and is preparing something, but has had to go off
for the rest of this afternoon, I think.

On Sat, 19 Sep 1998, John Local wrote:

> Dear all - can you provide me with a 'list' of your/our achievements/progress
> that I can incorporate into the presentation on Wednesday.
>
> The kinds of thing I'm thinking of are (not all necessarily true (at the
> moment):
>

>
> Data base recorded and labelled (what %?)
> Phonological parser built
> Lexicon parsed and tagged
> Hlsyn intelligent copysynthesis for perceptual tests

------- to clarify: for testing perception as in naturalness and
intelligibility; also for perception as in "whether we need to include
this bit of systematic variation" in the system; also for testing how much
desired spectral variation falls out of getting the rhythm right, and how
much more we need to build in. e.g. hypothesis: r/l resonances may fall
out of rhythm because they primarily reflect tongue body movement (maybe
lips tooo for /r/); whereas longer and darker onset /l/ before voiced
codas may require special accounting for. Thus PROCSY can be used for a
variety of perceptual tests, and as an acoustic front-end for Prosynth,
and hence as a test vehicle for Prosynth hypotheses, complemetary to
psola-manipulated concatenated speech. I'll send round a copy of the Heid
and Hawkins paper this afternoon. (NB: we can also use PROCSY to test the
level of deatil needed in the acoustic labelling. We're planning a system
where diffferent levels of detail can be plugged in, one idea being to
work out what level of deatil is NOT necessary e.g. , do we need to
mark/produce multiple bursts?)

> Temporal modelling begun (YorkTalk + first modelling from database)
> Analysis of co-articulatory effects (long-domain) begun

--- do you want details of these, or do you know them?

> Tools: xml search tools
> psola resynthesis tools specifically for testing hypotheses
> speech database labelled with acoustic events - syntactically and
> phonologically parsed

I think I already sent you a synopsis of the work Noel and I want to do.
Presumably that was sufficient, since it's not directly prosynth?
>
>
> In supplicant expectation.
>
> thanks
>
> John
>
> John Local
> Professor of Phonetics and Linguistics
> Department of Language and Linguistic Science
> University of York
> Heslington
> York Y01 5DD
>
> Tel 01904 432658
> E-mail lang4@york.ac.uk
> URL http://www.york.ac.uk/~lang4
>
>

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