Re: Specifying the audio buffer size

I think "latency" is a good name to use. Sounds like it would be a double
specifying seconds.

Ideally the actual latency measured would be from the time the audio/video
device receives the signal to the time the javascript sees it. Individual
implementations would probably have a minimum latency below which they
cannot operate, just like they support some bitrates or video resolutions
and not others.


On Tue, May 12, 2015 at 1:48 PM Harald Alvestrand <harald@alvestrand.no>
wrote:

> Den 11. mai 2015 22:28, skrev Jan-Ivar Bruaroey:
> > On 5/11/15 3:41 PM, Harald Alvestrand wrote:
> >> Den 11. mai 2015 17:08, skrev Jan-Ivar Bruaroey:
> >>>    { latency: { max: 0.0025 } } // I wanna do something else if not
> >>> low-latency
> >>>    { latency: { min: 0.025 } } // I wanna do something else if not
> >>> high-latency
> >>>
> >>> Lastly, constraints are about abstracting access to shared properties.
> >>> Since this use-case is ultimately about power consumption, I think it
> >>> fits (e.g. I might want to know if another tab prevents me from
> actually
> >>> getting high-latency, so I can alter my apps's behavior).
> >> This use case (speech recognition on mobile platforms) is about power
> >> consumption / latency tradeoff.
> >>
> >> The other use case I know of (interactive live music) is about seeing if
> >> delay can be controlled down to a tolerable level; power is almost
> >> irrelevant in this context.
> >
> > Makes sense. Being a constraint (vs. say a UA property) shouldn't
> > preclude a UA from in theory servicing such a request while
> > simultaneously servicing higher-latency requests at the same time, but
> > it wouldn't require a UA to do so, which I think works.
> >
> > .: Jan-Ivar :.
> >
>
> Yes, that's where I was coming from when I inserted "max" in the name -
> I think we should always allow lower latencies than were asked for -
> which contradicts the behavior of (say) width or height.
>
> But it is consistent with (say) bandwidth constraints, so I have no
> great beef about going with "latency" as the name of the constraint.
>
>