W3C home > Mailing lists > Public > www-style@w3.org > July 2011

Re: [css3-speech] voice-pitch and voice-pitch-range computed values

From: Daniel Weck <daniel.weck@gmail.com>
Date: Wed, 13 Jul 2011 21:42:28 +0100
Cc: fantasai <fantasai.lists@inkedblade.net>
Message-Id: <B760E501-634A-485C-BFF6-F51B62FCE541@gmail.com>
To: www style <www-style@w3.org>
I added an example to illustrate the point (scroll down a bit):

http://dev.w3.org/csswg/css3-speech/#voice-range

Let me know if this is what you had in mind.
(I think it is correct :)

On 13 Jul 2011, at 21:16, Daniel Weck wrote:

> I'm with you, thank you for the clarification / enlightenment.
> But we also need the computed value to allow absolute frequencies in addition to the keyword enumeration (the initial value is 'medium' but it may be overridden with, say, "200Hz"):
> 
> Computed value:
> an absolute frequency or keyword value, and potentially, a frequency, semitone, or percentage representing any non-zero offsets.
> 
> Does that sound right?
> 
> On 13 Jul 2011, at 19:47, fantasai wrote:
> 
>> http://dev.w3.org/csswg/css3-speech/#voice-props-voice-pitch
>> 
>> Value:          <frequency> && absolute? | <semitones> | <percentage> | x-low | low | medium | high | x-high
>> Initial:        medium
>> Applies to:     all elements
>> Inherited:      yes
>> Percentages:    refer to inherited value
>> Media:          speech
>> Computed value: specified value
>> 
>> "Specified value" isn't going to work because if I have
>> 
>> <parent>
>>   <child>
>>     <grandchild/>
>>   </child>
>> </parent>
>> 
>> parent {
>>   voice-pitch: low;
>> }
>> 
>> child {
>>   voice-pitch: 2st;
>> }
>> 
>> then child and grandchild should both have "low + 2st"; grandchild
>> should not be 2st higher than child.
>> 
>> The two technical options here are:
>> - compute an absolute frequency
>> - maintain a keyword and an offset
>> 
>> The difference in behavior would be discernable when the voice changes;
>> the latter would maintain the frequency of the old voice, whereas the
>> new one would recompute using the base frequency of the new voice.
>> 
>> The first behavior, however, is not useful: consider a switch from male
>> to female voice: if voice-pitch inherits as an absolute frequency, then
>> the pitch would be incorrect even in the simple case of 'voice-pitch'
>> never being specified.
>> 
>> So we have to use the second, perhaps something like
>> 
>> Computed value: a keyword and, potentially, a frequency, semitone, and/or
>>               percentage representing any non-zero offsets.
>> 
>> ~fantasai
>> 
> 
Received on Wednesday, 13 July 2011 20:43:07 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:42 GMT