can be handled automatically by "side-chaining".
does pulse know side-chaining?
My idea to implement this, by the way, would be to measure the perceptual loudness [3] of the current stream using Replay Gain [4] or similar. The resulting (instant) value would be used to set the volume for the notification stream. This can probably be all done inside PulseAudio by creating an appropriate module.
i wouldn't want main volume to change automatically..
this is a very individual thing that should be left to the user's individual preference/control..
Since I'm not an expert in signal processing, however, I don't know how difficult it would be to implement Replay Gain or a similar loudness measure in a way that can be used for this purpose. I also wonder what the impact on battery life would be. I'll try to look a bit more into these issues and report here when/if I find some answers. Of course, any useful pointers will be greatly appreciated.
you can't handle streams efficiently with Replay Gain.
Normalizing means the entire file has to be read first and scanned for transients..
Altering the dynamics of digital audio information would alter the information or message itself.. if you want to do this for flashmovies or the like, report the bug to Adobe.
you can't just simply alter the dynamics of a sound creation, without altering the actual message the respective sound is trying to convey..
dynamics are a major part of the semantics of sound.
normalizing a 3min audiofile takes quite a lot of CPU power and therefore doesn't make sense at this point in this context.
and yes: of course this kind of cpu load would shorten battery life, if constantly applied to files and streams when played..