Re: Timed Text roles

Hi all,

I am not too keen on having to pay $150 to access the CEA-708  
specification[1], so my knowledge of the "caption-text function  
tags" (section 8.5.9[2]) is somewhat limited. My understanding is  
that the proposed taxonomy is rather under-specified, and can  
essentially be summarized as 12 "roles" (please tell me if I'm wrong):
Dialogue, Source or speaker ID, Electronically-reproduced voice,  
Second-language dialogue, Voiceover, Dubbing, Subtitling, Voice  
quality, Song lyrics, Sound effects, Music description, Expletive.

Now, I appreciate that TTAF/DFXP aims at allowing interoperable  
interchange of symbols, but I also believe that the lack of  
interoperable semantics will confuse implementors, to say the least.  
In my opinion, the list of roles defined in section 12.2.2[3] have no  
reason to exist if they are "meaningless" (i.e. without semantics).  
They should either be removed, or namespace-qualified to point to the  
CEA definition (if any).

Within W3C, it seems to be common practice to define a minimum  
taxonomy for semantic roles: the XHTML-role module provides a small  
list of default symbols with a simple plain-text description[4], and  
a more structured RDF/OWL taxonomy[5].

Another approach is to define a taxonomy in a separate document, like  
WAI-ARIA Roles[6]. Actually, this is an interesting example, because  
the "dialog" role defined in TTAF/DFXP conflicts with the "dialog"  
role[7] in WAI-ARIA Roles !

So far, the TT-WG has identified the following list of default roles  
(of which quite a few are ambiguous to me):
"action", "caption", "dialog", "expletive", "kinesic", "lyrics",  
"music", "narration", "quality", "sound", "source", "suppressed",  
"reproduction", "thought", "title", "transcription".

Although semantic roles do not enforce specific user-agent behaviour,  
as an implementor I would like a clarification to make sure that what  
I understand is also what other people think.

Kind regards, Daniel Weck.

[1] http://www.ce.org/Standards/StandardDetails.aspx? 
Id=1782&number=CEA-708-C
[2] http://www.ce.org/PDF/PREVIEW__pages_from_CEA-708-C_FINAL.pdf
[3] http://www.w3.org/TR/ttaf1-dfxp/#metadata-attribute-role
[4] http://www.w3.org/TR/xhtml-role/#s_role_module_attributes
[5] http://www.w3.org/TR/xhtml-role/#s_rdf
[6] http://www.w3.org/TR/aria-role/#roles
[7] http://www.w3.org/TR/aria-role/#dialog

On 11 Oct 2007, at 14:49, Glenn A. Adams wrote:
> The roles are, at present, intended to be "standard" only in the sense
> that there is a "stsandard" set of keywords. What those keywords  
> mean is
> up to the author(s). I agree that this does not permit interoperable
> interchange of semantics, but it does permit interoperable interchange
> of symbols as such. Just think about what the <strong/> or <em/> tags
> mean in HTML. How would you define them in a meaningful way?
>
> It would be going overboard to recommend that authors NOT use these
> keywords if they can find a meaningful use for them. If some group of
> authors and some group of recipients should wish to define a profile
> that provides further semantics or context, then there is nothing to
> prevent doing this. Many standards define facilities that for  
> effective
> interchange requires profiling.
>
> However, if you would like to propose a set of definitions for
> standardization, then please feel free to do so, and I will make sure
> that the TTWG considers them. But even then, how would you enforce
> consistent use with such definitions?
>
> Regards,
> Glenn
>
>> -----Original Message-----
>> From: Benjamin Hawkes-Lewis [mailto:bhawkeslewis@googlemail.com]
>> Sent: Thursday, October 11, 2007 8:32 PM
>> To: Glenn A. Adams
>> Cc: public-tt@w3.org; w3c-wai-ig@w3.org
>> Subject: Re: Timed Text roles
>>
>> Glenn A. Adams wrote:
>>> The TTWG discussed this need, and concluded it would be exceedingly
>>> difficult or impossible to define them in a way that would retain
> their
>>> utility while not overly constraining such use.
>>
>> Thanks kindly for the response, but I don't understand at all. :(
> Could
>> you please explain the WG's reasoning a bit more? Given that authors
> can
>> always create a new role in the x- prefixed space, how would defining
>> the standard roles constrain authoring? What is the real-world  
>> utility
>> of a standardised set of roles that aren't defined in any way? It
> seems
>> to me that not defining "transcription", for example, means that TT
>> fails to provide "clear indications in the format of what text
>> corresponds to speech in some corresponding audio segment" as
> requested
>> by Alfred S. Gilman:
>>
>> http://lists.w3.org/Archives/Public/public-tt/2005Apr/0043.html
>>
>>> It was recognized that
>>> some existing standards, such as the US Digital Television Closed
>>> Captioning (DTVCC) standard, CEA-708, similarly enumerated such a
> list
>>> but without providing any further definition.
>>
>> Are the standard roles precisely the same as CEA-708? If they are,  
>> why
>> doesn't the TT spec say they are? And if they aren't, then what is  
>> the
>> relevance of this apparent anti-pattern in CEA-708?
>>
>> Might it not be better to specify cea- prefixed roles for mapping
>> CEA-708 roles to Timed Text?
>>
>>> Notwithstanding the above, we may consider adding informative
> examples
>>> to the text of DFXP during the process of transitioning from CR to
> REC.
>>> If you would like to submit such examples, possibly with
> descriptions,
>>> then that would be most welcome.
>>
>> Well, I'd be happy to do that, but I can't submit examples or
>> descriptions when I can't tell what the roles are for! If existing
> data
>> in CEA-708 form is to be mapped to such roles, then we need to know
> how
>> such labels are currently used by people who use CEA-708: I can't  
>> just
>> make things up. And if they use them in utterly incoherent ways, then
>> shouldn't a new format provide a set of coherent roles?
>>
>> As things stand, I would have to recommend that authors don't make  
>> any
>> use of these ambiguous standard roles, but instead publish some sort
> of
>> microformat using the x- space. At least then one could find out what
>> the author intended.
>>
>> Has the WG considered allowing people to specify a URI that defines
> the
>> the roles they are using so that there is no possibility of  
>> confusion?
>> Compare the approaches of HTML 4.01, the draft XHTML 2.0 Role
> Attribute
>> Module, and GRDDL:
>>
>> http://www.w3.org/TR/html401/struct/global.html#profiles
>>
>>
> http://www.w3.org/TR/xhtml2/mod- 
> roleAttribute.html#s_roleAttributemodule
>>
>> http://www.w3.org/TR/grddl/
>>
>> (PS Apologies to w3c-wai-ig for the noise. I mistook public-tt latest
>> for the whole public-tt archives: turns out that thankfully there are
>> people on this list.)
>>
>> --
>> Benjamin Hawkes-Lewis

Received on Thursday, 11 October 2007 15:39:07 UTC