Re: [css3-speech] square brackets group

On 10 Dec 2010, at 23:51, Daniel Weck wrote:
> I am questioning the use of the <number> discriminator based on a  
> <specific-voice>. I can understand the selection logic based on a  
> <generic-voice>, but I am not sure how the "preferred variant"  
> heuristics can apply to a specific voice instance.


After further exploration of the SSML specification, I changed my mind  
on this particular point (context given in the full quote below). For  
example, there may be several *male* and *child* voices named *Mike*  
on a given system, so the "voice variant" mechanism described by SSML  
applies identically in CSS-Speech, based on an index that indicates a  
preferred alternative (starting from 1). I therefore fixed the grammar  
so that the <number> discriminator applies to the whole set of voice  
criteria.

Note that the initial comment about misplaced grouping indicators  
still stands, and has been fixed appropriately.

Regards, Daniel


On 10 Dec 2010, at 23:51, Daniel Weck wrote:
> On 10 Dec 2010, at 23:09, Bjoern Hoehrmann wrote:
>> Yes, but if you consider square brackets to represent grouping,  
>> then the
>> examples in the draft do not match the grammar, `voice-family: male  
>> 2`
>> for instance is not valid because <age> is not optional but omitted  
>> in
>> the example. However, it would match if you read (some of the) square
>> brackets to represent optionality. I do agree that the Working Group
>> does not seem to have addressed my comment yet.
>
>
> The <age> and <number> fields of the "voice-family" CSS 3 Speech  
> property [1] were added to the fields specified by CSS 2.1 Aural  
> style sheets [2].
>
> There seems to be a typo in the CSS 3 grammar indeed, as the  
> examples are correct (they are backward compatible with the CSS 2.1  
> notation).
>
> I would suggest the following errata (line breaks for clarity, "xxx"  
> is the exact duplicate of line #2):
>
> ----------------
> [xxx,]*
> [[<specific-voice> | [<age>? <generic-voice>]] <number>?]
> | inherit
> ----------------
>
> ...but actually I am questioning the use of the <number>  
> discriminator based on a <specific-voice>. I can understand the  
> selection logic based on a <generic-voice>, but I am not sure how  
> the "preferred variant" heuristics can apply to a specific voice  
> instance. Therefore I hereby present the errata of the errata:
>
> ----------------
> [xxx,]*
> [<specific-voice> | [<age>? <generic-voice> <number>?]]
> | inherit
> ----------------
>
> Comments welcome.
> Regards, Daniel
>
> [1]
> http://www.w3.org/TR/css3-speech/#voice-family
>
> [2]
> http://www.w3.org/TR/CSS21/aural.html#propdef-voice-family

Daniel Weck
daniel.weck@gmail.com

Received on Saturday, 22 January 2011 12:16:47 UTC