Re: User media constraint: peerIdentity

On 10/26/2012 02:58 PM, Timothy B. Terriberry wrote:
> Harald Alvestrand wrote:
>>> The bigger problem, to my mind, is that it is now impossible to
>>> implement a level meter for the audio, so that people can check their
>>> mic. We can still provide one in browser chrome in gUM, but not after
>>> permission has been granted.
>>>
>> yes, it's now impossible to get an audio volume meter by channeling the
>> audio into the WebAudio API.
>>
>> If we implement an "audio volume" statistic via the stats API, which has
>> been suggested, there is an alternate path; I don't think statistics are
>> prohibited for tainted media.
>
> I was thinking more along the lines of providing an HTMLLevelMeter
> element, which would have the benefit of also being useful to developers
> who don't want to go through the difficulty of using the WebAudio API to
> compute mic levels manually. Though I'd be happy to delay such a thing
> until "version 2".

I think that this sounds compelling. And given that we have this sorted 
for the non-tainted case (using web audio API), perhaps delaying it 
would not be a big problem (though this is a central piece of 
functionality IMO; I use the volume meter all the time to check that the 
right mike is used).

>
> The difficulty with exposing this via statistics is that now you've
> created a side-channel leaking information about the tainted streams
> with, I expect, roughly as much information about them as the much-hyped
> VBR information leak.
>

Received on Friday, 26 October 2012 13:10:21 UTC