RE: Vowel Epenthesis and Audiograms

We've found that the screen reader population enjoys
high speed, low prosody output. They are really forced
to be in this category in my sighted person's opinion,
simply because of the un-navigability of the medium.
We commonly skip around when we read, but how can you
code that?


there have been many papers on this subject.
I would refer you to Silke Goronzy, Lecture notes in computer
science, vol. 2560, Jan 2002 p. 79 for adapting
the recognition task to the matched prosodic
output. Accent can be recognized. I think the attribute of prosody
should be divided into subtypes if its going to be a valid
attribute (wow, my sgml is coming back to me!)


Once we get the spelling-it-all-out type work done,
there should be some kind of reference to make
this hang together.

I agree the phonemic
approach is the right way.



-----Original Message-----
From: www-voice-request@w3.org [mailto:www-voice-request@w3.org]On
Behalf Of Max Froumentin
Sent: Wednesday, March 09, 2005 8:49 AM
To: Jim Tobias
Cc: 'Al Gilman'; www-voice@w3.org; 'Harvey Bingham'
Subject: Re: Vowel Epenthesis and Audiograms



"Jim Tobias" <tobias@inclusive.com> writes:

> There is a clear potential benefit for people who are hard of hearing or
in
> noisy environments, but this may be even more valuable when the speech
rate
> is set high, such as by screen reader users.

Sounds interesting, and it gives me two ideas:

- there could be an attribute on the grapheme element, indicating that this
  particular grapheme contains a more accessible pronunciation:

  <lexeme>
    <grapheme>huge</grapheme>
    <phoneme>hju:dz</phoneme>
    <phoneme class="articulated">hju:u:u:u:dzzzz</phoneme>
  </lexeme>

- or why not have some SSML markup (prosody, for instance) there too?

  <phoneme class="slow"><prosody rate="-10%">huge</prosody></phoneme>

  maybe _that_'s streching the thread out far...

Max.

Received on Thursday, 10 March 2005 01:47:19 UTC