Re: Stats variables - some more input

Harald Alvestrand wrote:
> One question that arose about audio volume representation: It's hairy.
> Especially it's hairy to define volume in a way that makes sense when
> you don't know the sampling interval (because this is a polling
> interface), and you don't want to give undue importance to the last
> tenth of a before the sample. Or perhaps you do - it might be app
> dependent.
>
> One suggestion made internally: Just report the cumulative audio energy
> (sum of audio energy over time) instead. Then people can transform that
> as they will.
>
> Thoughts on this?

My recommendation would be to follow EBU R128: 
<http://tech.ebu.ch/loudness>. See specifically EBU Tech Doc 3344.

Received on Tuesday, 12 November 2013 07:58:27 UTC