[css3-speech] Comments on WD 20040727

I've not got any experience with voice technologies per se, but these were 
the questions that immediately occurred to me while I was reading the latest 
draft: 

* For 'voice-volume' and 'cue', I understand that the named volumes ('soft', 
etc) don't have pre-defined mappings into the 0-100 range, where 0 is silent 
and 100 is the loudest value (and the value set by the user). However, 
there's no indication about what value is the 'preferred' volume (which is 
what 'medium' is, as the default). It can't be 100, since that would mean 
that both 'loud' and 'x-loud' would have no effect, so it must be some other 
number. 

I can see that you don't want to define a number, since if you were to 
define it to be 75, for example, that would force the maximum amplitude to 
be 100/75 = 33% higher than the preferred volume, by definition. 

However, without a definition of what the preferred value is, I can't see 
that the <number> value is at all useful, except for voice-volume:0 and 
voice-volume:100. Since a UA will have to make a choice as to the mapped 
value of 'medium', the only result of not defining it would seem to be 
either that the property (specified as a <number>) becomes non-interoperable 
(rendering it useless), or that all implementations tend to the same default 
(as for font-size). 

* Presumably, voice-volume: silent should be equivalent to voice-volume: 0. 
This is implied, but not defined. 

* It should be explicitly mentioned that voice-volume: medium represents the 
user's preferred volume. 

* Although 'voice-volume' "does not apply to cues", it does presumably set 
the default volume if the cue does not override it (see next point). This 
should be clarified, as the current text implies there is no connection. 

* 'cue', 'cue-before', and 'cue-after' do not define what volume the cue 
will be played at if the amplitude is not specified. Presumably, it's as if 
'100%' was used - i.e., the inherited voice-volume. 

* When a percentage is specified for 'cue', should the value not be derived 
from the computed value of 'voice-volume' rather than the inherited value? 
For example, what would be the expected volume of the following? 

{
 voice-volume: loud;
 cue: url("pop.au") 100%;
} 

Should the computed volume for the cue be equal to 'loud', or to the 
inherited value of voice-volume? 

* Why is the set of named volume levels for 'cue' not defined to be 
identical in amplitude to the set defined for 'voice-volume'? 

* "The following two rules are equivalent: 

div {mark-before: start; mark-after: end }
div {mark: begin end }" 

I assume the latter should be 'mark: start end'? 

* 'voice-level' is mentioned twice and documented nowhere. Is it a typo for 
'voice-volume'? 

* Will 'voice-family: male 1' always be the same as 'voice-family: male'? 
This isn't defined. Will the 'next' voice always be 'male 2', and so on, 
using consecutive numbering? 

* Why is the initial value of 'voice-rate' not 'medium', defined to be 
equivalent to 100%? This is the same problem as my first point, except that 
this time, the named values are the ones that aren't defined. 

* Why are percentage values for 'voice-rate' relative to 'the default rate' 
(so that 100% is the default value), but percentage values for 'voice-pitch' 
and 'voice-pitch-range' are relative to the inherited value? Presumably, 
there's some rationale behind the decision. 

Regards,
Malcolm

Received on Tuesday, 27 July 2004 13:45:02 UTC