W3C home > Mailing lists > Public > www-style@w3.org > May 2004

CSS3 Speech module: Comments and Corrections

From: Nicholas Shanks <contact@nickshanks.com>
Date: Fri, 7 May 2004 14:32:54 +0100
Message-Id: <0B882B61-A02B-11D8-80CF-0005020F149E@nickshanks.com>
To: www-style@w3.org

After having read through the new aural media WD (CSS3 Speech module, 
http://www.w3.org/TR/css3-speech/), I would like to offer the following 
comments to the author (which isn't given, BTW) and make a small 
correction. Apologies if these had been brought up before but archive 
searches turned up nothing.

1) There appears to be no direct way to substitute an audio file in 
place of a specific span of text, which would not be rendered by the 
UA. For example, the closest I can get (using 'workaround' markup) is:

HTML:
<q id="lifeistough"><span class="clip" />Life is tough, but it's 
tougher when you're stupid.</q>

CSS:
q#lifeistough { speak: none; }
q#lifeistough .clip { speak: normal; cue-before: url(johnwayne.aiff); }

Is this the only/best way? May I ask that a more obvious mechanism be 
added to the recommendation, or that such an example be included in the 
documentation.

2) Corrections to and discussion on the language properties of content.
The fragment at http://www.w3.org/TR/css3-speech/#voice-char-props 
erroneously states that the country value denoting Britain for the 
xml:lang attribute is "en_uk" when in fact it is "en_gb". Furthermore 
it then states that this doesn't help with in-country variations such 
as scottish or welsh. There is a commonly cited example in HTML 
tutorials of using "en-cockney" as a language-region identifier for 
HTML pages, I would offer that "en-scots" and "en-welsh" would be valid 
specifiers for the cases you have provided, though I don't know the 
origins of this syntax. In such a case it would be up to the user agent 
to realise that 'scots' is a dialect of the region 'gb', and to 
substitute an 'en-gb' voice when 'en-scots' is unavailable. Perhaps a 
work-around for this would be using multiple levels of region 
specificity, such as "en-gb-scots". Comments?

On the issue of languages spoken with foreign accents, as the accent is 
a rendering style, and the language is an intrinsic property of the 
content, I would offer that xml:lang be used to specify the language 
content of the text, and an additional voice-accent property be created 
(or an accent property instance be added to voice-family) taking a 
language-region specifier, for example:

HTML:
<span class="sexymexicanmaid" xml:lang="en-us">Toast and marmalade 
<span xml:lang="es-mx">señor</span>?</span>
-or-
<span class="sexymexicanmaid" xml:lang="es-mx">¿<span 
xml:lang="en-us">Toast and marmalade</span> señor?</span>

CSS:
.sexymexicanmaid { voice-family: young female; voice-accent: es-mx; }

Which should render the whole question in a mexican accent, with the 
languages correctly marked up to boot.
Again, comments are welcomed by anyone wishing to contribute.


- Nick.
Received on Friday, 7 May 2004 09:52:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 27 April 2009 13:54:29 GMT