Re: [css3-speech] cue volume

I totally agree with you. My original modification was to map cue  
volume relatively to voice-volume (not just for the 'silent' case).  
The problem was in describing this desired effect in a implementable  
way (i.e. intrinsic audio cue  average wave amplitude isn't  
necessarily the same as standard voice volume level from TTS engine).  
Let me have another go now that we have improved voice-volume. We can  
completely make the "volume level" solution rely on the user-perceived  
loudness, i.e. on the keywords. Authors don't need to care about  
intrinsic volume levels, which should be normalized/handled by the  
user-agent in order to offer a decent listening experience.

On 6 Jul 2011, at 23:33, fantasai wrote:

> On 07/06/2011 03:14 PM, Daniel Weck wrote:
>> On 6 Jul 2011, at 22:56, fantasai wrote:
>>> I'm wondering now, voice-volume controls the cue when
>>> it's value is 'silent', but not when it's anything else. That seems
>>> somewhat odd. Do we want cues linked to the voice-volume? Or should
>>> they be independent? (It is *voice*-volume, after all.)
>> We talked about this before: the problem is that in the "aural box
>> model", the selected element is "decorated" with pause, rest and cue,
>> so when the element itself is silent (well, its contents), its cues
>> should be too (there is an analogy with margin, padding, and border
>> => the element visibility impacts its "decorations" too).
> Makes sense. Should shifts in the voice volume also affect cues?
> Or optionally affect cues (e.g. if there's an 'auto' cue volume
> then the difference between the voice-volume and 'medium' is applied
> to the cue)?
> Also how would cues be aligned with the voice volume, anyway? This
> does seem to be important. Like, suppose my device is reading the
> page in a loud environment. I've set my preferred volume ('medium')
> to be loud, but the cues are all pre-recorded, so they will not be
> loud. Then suppose I'm using my reader in a quiet environment. I
> set the voice volume to be quiet. But the cues will still be loud.
> Seems like a problem.

Received on Thursday, 7 July 2011 00:01:32 UTC