Re: Fwd: Re: Audio - In kernel audio mixing

[Nathanial Sloss <nat%netbsd.org@localhost>]

Hi Brian,
On Tue, 17 May 2016 03:04:32 Brian Buhrow wrote:
> 	hello.  I'll echo Brad's questions about latency and delay.  Delay,
> when it comes to audio presentation, is bad.  I was recently at a talk
> where the developer was discussing trying to synchronize one audio source
> with another.  For their application, they're sending audio sources from
> two locations over WiFi and other wireless technologies.  Their goal was to
> make sure the sychronized audio was never more than 25ms out of sync, and,
> hopefully, a lot less.  If my latency budget is only 50-100ms for a given
> audio application, I'd hate to spend so much of that budget before I even
> get out of the kernel.  I don't know what the latency is for the
> traditional audio drivers is, but I can tell you it varies depending on the
> driver in question.  For example, the auich(4) and emuxki(4) drivers are
> very responsive with low latency.  However, the uaudio(4) driver exhibits
> considerably more delay in its operation.  If these vaudio drivers are
> slower than the uaudio(4) driver, then I'd suggest working to get that
> delay down.  If that's just not feasible, then the traditional audio(4)
> interface should remain viable for those of us who need very responsive
> audio performance.
> 
> -thanks
> -Brian


Please check out ftp.netbsd.org/pub/NetBSD/misc/nat/nexaudio3-kern.diff as it 
introduces no additional latency or delay.

Best regards,

Nat.