Re: [css3-speech] voice-balance and azimuth

On 07/06/2011 01:49 PM, Daniel Weck wrote:
>
> On 30 Jun 2011, at 01:08, fantasai wrote:
>
>> # The ‘voice-balance’ property manipulates the distribution of audio
>> # output between left and right channels in stereo-capable sound devices.
>>
>> So.. I'm a little concerned about this definition, as we can't have
>> both a 'voice-balance' property /and/ an 'azimuth' property, since
>> afaict they control essentially the same thing. I think it's fair to
>> assume that we'll at some point add controls for surround-sound and
>> design CSS with that in mind, so that we don't paint ourselves into
>> a corner here.
>>
>> So, if we wanted to add 'azimuth's capabilities to CSS how would that
>> work? Do we want to have the current definition as well as <angle>
>> combined into 'voice-balance'? Revert <number> to <angle> and define
>> how it maps into 2-channel audio? Something else?
>
> I am not aware of any existing CSS speech implementation of basic
> left/right balance, let alone of proper "spatial audio"
> (azimuth and elevation). So I think we have plenty of flexibility to
> change the current draft, and to diverge slightly from Aural CSS2.1.
>
> The current definition of 'voice-balance' is:
>
> <number> | left | center | right | leftwards | rightwards
>
> Let's assume that the CSS Speech Level 3 Module ships only with simple
> stereo balance capabilities, and let's imagine that Level 4 eventually
> supports proper "spatial audio" by adding the following values for
> 'voice-balance' (the goal is to ensure that Level 3 content still works
> with Level 4 user agent implementations):
>
> <angle>
> left-side, right-side
> center-left, center-right
> far-left, far-right
> (+behind)
>
> The old <number> values [-100, 100] can be mapped to equivalent <angle>
> values, using a simple mathematical formula.

Ok, how about we put this formula in CSS3 Speech then, so that a
surround sound system will know what to do with these values.

> The old 'leftwards' keyword value would still apply (with similar
> semantics), but unlike CSS 2.1, I would propose to define a separate
> "turn counter-clockwise" value for Level 4 ('leftwards' doesn't make
> sense when reaching angular values that actually
> "point" rightwards!). Same remark for 'rightwards'.

I agree 100% on this point. :)

> The old 'center' keyword retains the same semantics (and the 'behind'
> keyword can be combined in Level4 to refer to, well, behind the listener
> instead of in front of him/her).
>
> The old 'right' and 'left' keywords could actually still match the new
> definitions. For example, the transition from full frontal (0 degrees)
> to full right (90 degrees) is:
> center -> center-right -> right -> far-right -> right-side
> As you can see, 'right' is in the "middle" of the transition, which is
> perfectly acceptable (it matches the user experience with CSS Speech
> Level 3).
>
> So to conclude, I don't think we need to change anything to the current
> Level 3 draft :)

One thing you haven't addressed is the distinction between balance and
direction. In a stereo system, these two things are the same control.
But in a surround-sound system, they are different.

For example, if I output the same amount of sound from all speakers in
the system, on a stereo system this is the same as a center-located
voice, but in a surround-sound system, a center-located voice would
instead be sound coming out of only the front-center speaker.

~fantasai

Received on Wednesday, 6 July 2011 21:44:13 UTC