Now that we have
some more or less clear scenarios, the first design step should be defining a
list user needs, before we think of any possible controls to cover them. These are the ones I've identified:
* Current application is too loud -> only emergency case, there should be a quick access to a function that mutes all sound (and a way to limit volume before deactivating the mute!).
* Refine volume in a multimedia stream. Only case that really needs fine control, IMHO.
* Make sure that any possible sound is not louder than X decibels (think "late at night"). Only case where a global setting is really needed. A slider doesn't guarantee this, by the way - this should work as a global cap on sound.
* Make a system sound (warnings, interface feedback) louder or softer than another kind of sound. Best fit should be a priority based control - but not necessarily a list of all applications, more like a temporary "muffle this sound source".
All scenarios described in previous posts (presentation mode, mixing several streams�) are particular instances or combinations of these needs.
We should aim to provide an interface that satisfy these user goals in an easy way, so that the user has complete control over the things that really matter. This way she can tweak them at her will in ways that none of us could think of in advance.
I don't think a volume slider per application is a good way to satisfy those goals. If only because after the task ends and the goal is accomplished you're forced to manually restore all the controls to its previous state, one by one.
Can you think of other needs not listed above, and of ways to represent them with direct manipulation widgets? I have some ideas, but first want to hear your opinions about this analysis.