I think sending realtime audio data via eternet will be
extremely difficult. It should be impossible to synchronize
sample clock via ethernet. So each device has to have its
own sample clock. This requires:
- resampling units (=no lossles connections possible)
or
- large buffers to slowly adapt s
Behalf Of stef
> Sent: 18 February 2002 17:01
> To: [EMAIL PROTECTED]
> Subject: Re: [Alsa-devel] audio via ethernet (was: dolby digital output)
>
>
> I think sending realtime audio data via eternet will be
> extremely difficult. It should be impossible to synchronize
> sample c
A parallel discussion has been taking place in the music bar
([EMAIL PROTECTED]). I've given some of the more important comments
below.
>I think sending realtime audio data via eternet will be
>extremely difficult. It should be impossible to synchronize
>sample clock via ethernet. So each device
On Mon, 18 Feb 2002, stef wrote:
> I think sending realtime audio data via eternet will be
> extremely difficult.
Why?
> It should be impossible to synchronize
> sample clock via ethernet.
Why?
Did you read the gibson magic spec?
-Dan
--
[-] Omae no subete no kichi wa ore no mono da. [-]
_
Dan Hollis wrote:
>
> On Mon, 18 Feb 2002, stef wrote:
> > I think sending realtime audio data via eternet will be
> > extremely difficult.
>
> Why?
>
> > It should be impossible to synchronize
> > sample clock via ethernet.
>
> Why?
>
> Did you read the gibson magic spec?
Oh, sorry! I thoug
On Mon, 2002-02-18 at 17:00, stef wrote:
> It should be impossible to synchronize
> sample clock via ethernet.
If it's possible for me to synchronise my PC clock to sub-millisecond
accuracy over *IP* on a LAN, I somehow doubt it'd be impossible to
synchronise word clocks over ethernet.
--
Bob H
Bob Ham wrote:
> If it's possible for me to synchronise my PC clock to sub-millisecond
> accuracy over *IP* on a LAN, I somehow doubt it'd be impossible to
> synchronise word clocks over ethernet.
Sounds great! 20 nanoseconds would be enough.
But the problem in digital audio is that the clock h
On Tue, 19 Feb 2002, stef wrote:
> Anyway, which software/protocol/service do you use for
> your clock synchronisation? _very_ interesting!
Probably ntp. Works well, multicast mode should be excellent for this.
BTW if you use fixed size packets on fullduplex ethernet (as gibson's
magic does), t
On Mon, 2002-02-18 at 23:00, stef wrote:
> Anyway, which software/protocol/service do you use for
> your clock synchronisation? _very_ interesting!
The mother of all time protocols, NTP
http://www.eecis.udel.edu/~mills/ntp.htm
--
Bob Ham: [EMAIL PROTECTED]
My music: http://mp3.com/obelisk_uk
On Tue, 19 Feb 2002, stef wrote:
> But the problem in digital audio is that the clock has
> to be readjusted in short intervals.
I just thought of something. multicast is perfect for clock
synchronization.
-Dan
--
[-] Omae no subete no kichi wa ore no mono da. [-]
___
uary 2002 23:01
> To: [EMAIL PROTECTED]
> Subject: Re: [Alsa-devel] audio via ethernet (was: dolby digital output)
>
>
> Bob Ham wrote:
>
> > If it's possible for me to synchronise my PC clock to sub-millisecond
> > accuracy over *IP* on a LAN, I somehow doubt it
On Mon, Feb 18, 2002 at 03:11:11 -0800, Dan Hollis wrote:
> On Tue, 19 Feb 2002, stef wrote:
> > Anyway, which software/protocol/service do you use for
> > your clock synchronisation? _very_ interesting!
>
> Probably ntp. Works well, multicast mode should be excellent for this.
It's not reasonab
This one is a toughy. Even if you use NTP, you still have a signifficant
amount of drift to deal with. The system clock is not
sample-accurate. Multicast NTP still requires unicast connectivity to the
multicast servers, so I really don't understand the point.
_J
In the new year, Steve Harris
On Tue, 19 Feb 2002, Steve Harris wrote:
> However, on raw ethernet, the arrival of the packets + a timestamp should
> be enough synchronisation.
Yes, it should.
> >From what I remeber of magic, it ensures the packets are sent
> synchronosly.
No, it just tries very hard :-) by using fixed packe
On Tue, Feb 19, 2002 at 01:48:48 -0800, Dan Hollis wrote:
> > >From what I remeber of magic, it ensures the packets are sent
> > synchronosly.
>
> No, it just tries very hard :-) by using fixed packet lengths, and strict
> inter-packet timings.
;) OK, but as long as the system is magic only the
15 matches
Mail list logo