Re: WAI Action items

On Fri, 13 Jun 1997 jim@arkenstone.org wrote:

> 
> OK, Daniel, I'm operating on the assumption that I'll be
> working the phonetic pronunciation issue.  I'll make the
> June 18 phone call on the assumption that timeliness issues
> affect this effort.

That is good to hear. Although I have been largely responsible to
raising this issue, my time is mostly occupied with work on HTML.

Perhaps its worth my summarising the points I am aware of:

The need is for a way to encode phonemic and prosodic information
for irregular words and phrases that are unlikely to appear in
dictionaries, e.g. people's names. I envisage dictionaries for
different languages, being selected upon the basis of the the
HTML Cougar LANG attribute which uses RFC 1766. 

The encoding should be easy enough for people to make a reasonable
job of their own names without needing to be experts in digital
speech. This suggests the use of ASCII encodings based up on
letter sequences that suggest the intended pronunciation. Note
that this introduces cultural biases, so that we may need to
think about multiple "input" syntaxes, and a common back-end
syntax as is being developed by the W3C HTML-Math working group.

The encoding should however be capable in expert hands of more
faithful/subtle behaviour. In addition, it shouldn't adversely
effect the ability to run speech synthesisers fast, since people
are used to doing this to scan texts quickly. 

The advent of software based speech synthesisers will largely
obviate some of the difficulties with today's hardware-based
systems. This gives us greater freedom to choose the kinds of
features we want, without the constraints imposed by a policy
of lowest common denominator for different hardware-based
implementations.

Regards,

Dave Raggett <dsr@w3.org>

tel +44 122 578 2521 url http://www.w3.org/People/Raggett
World Wide Web Consortium (on assignment from HP Labs)

Received on Saturday, 14 June 1997 17:52:05 UTC