Vote for a MM system (Was: Re: summary of the aKademy meetings)

[Lennart Poettering]

On Thu, 09.09.04 16:05, Allan Sandfeld Jensen (kde at carewolf.com) wrote:

> I am sorry if I missed something. I only took a quick glance at polypaudio to 
> see if you fixed all of the deficiencies of esd, and synchronizations wasn't 
> advertised on the front-page, but I will have a closer look now.
> 
> Actually I would expect networked audio over a high latency network to work 
> fine, as long as the buffersize matched the fluctuations in bandwidth. For 
> instance if you use 200ms of buffer on the recieving side, and advertise that 
> to the client-streamer. Everything could be synchronized fine, as long as no 
> more than 200ms of audio is ever in transit at any time. If that happend you 
> would need to increase buffersize and resynchronize.

That part isn't a problem. The issue is synchronization, meaning: when
will the sample I write to the sound server now be played? It is
impossible to solve this problem precisely. See: you're playing a
movie on machine A using a sound card on machine B. To synchronize
video and audio the movie player needs to know when to show the right
video frame for the a certain sound sample. So it queries machine B:
"what is your current latency?". This request takes some time to
travel to machine B, let's say 10ms. Machine B replies with
"50ms". That reply takes some time to travel back to machine A. let's
say 5ms. The real latency is 50ms+10ms. but machine A can only
estimate that the real latency is something like 50ms+7.5ms. (7.5ms is
half the roundtrip time.)

Lennart

-- 
name { Lennart Poettering } loc { Hamburg - Germany }
mail { mzft (at) 0pointer (dot) de } gpg { 1A015CC4 }  
www { http://0pointer.de/lennart/ } icq# { 11060553 }