[ardour-users] Ardour and Latency Handling

[Andrew Johnston]

On 5/1/06, Wolfgang Woehl <tito at rumford.de> wrote:
>
> Sunday 30 April 2006 06:52, Andrew Johnston:
>
> > on the #ardour irc channel last night, I was informed that for standard
> > multitrack recording and overdubbing, latency does not factor into the
> > situation if not using softsynths, or live dsp performing as ardour
> handles
> > the latency internally, and syncs all the tracks to line up with each
> > other, based on the latencies reported to it by jack and by the ladspa
> > plugins.
>
> Couldn't say it better. It works like that. What's your Ardour version,
> jackd
> version?
>
> > I have an RME HDSP9652 connected via adat to a Behringer DDX3216.  I
> routed
>
> Have you measured the DDX3216's latency? It'll have some and you can add
> those
> numbers to your jackd startup options with -I and -O  -- if your external
> routing always includes the DDX.
>
> > the click track out channel one, and on my mixer rerouted channel one
> input
> > back out to channel 2.  I then recorded channel 2 in ardour and
> therefore
> > got the click track recording at the same time as playing.  I then
> played
> > back from the beginning with click track still enabled.  At -p 64 there
> was
>
> To choose Ardour live click for loopback tests is not the best choice
> because
> click is special. That is unfortunate but non-trivial to change I think.
> Record the click to a track and start from there.
>
> > no delay between the two.  At -p 1024 there was a noticeable delay.  I
> went
>
> When testing this stuff I find it really helpful to have some actual
> numbers
> like audio frames. More telling than "noticeable". Right-click the clocks
> to
> change their format and write down the delta between edit cursor and
> playhead
> for example. Looking at the numbers can give you a clue about what's going
> on. Remember to measure the latency of your external gear.
>
> > one step further and switched off the click, and used the exisiting
> > recorded material (now it's own click track) and routed it out and back
> in
> > on another track.  I returned the same results, and regardless of the
> > toggling the alignment options, there was no difference.  This leads me
> to
> > believe that in fact ardour does not handle latency internally and
>
> Ardour does compensate for latency. With the -I and -O numbers in the
> picture
> you get sample-perfect alignment.
>
> I wonder what's mixed up in your setup. On connecting tracks to sources
> Ardour
> asks JACK what kind of source that is: Hardware or software source. And it
> sets up alignment of the track accordingly. This can get messed up when
> you
> change connections later on. This is still buggy in Ardour.
>
> --
> Wolfgang
> _______________________________________________
> ardour-users mailing list
> ardour-users at lists.ardour.org
> http://lists.ardour.org/listinfo.cgi/ardour-users-ardour.org
>


Thanks for the replies.

Well, I did some more testing and still coming up with the same results
FYI, I;m using a gentoo 64bit system.
Ardour is 0.99.2
Jackd is 0.99.0-r1

I did a latency test on my console using jdelay and it came up at about
1/10th of a millisecond, taking the JACK delay into account (assuming a read
the figures correctly).
So the problem is not the mixer.  I checked for samples and have come up
with some more useful information than before.  Zooming in close I can see
ardour has corrected for latency, but just not correctly.  For example with
Jack set to say 2048 frames, the track being recorded is placed 2048 frames
BEFORE the original track.  With 64 frames, it is 64 frames before, but
because this is only 1.3 milliseconds, it is not noticeable as delay.  I
have tried changing the JACK latency option within ardour, just to test, but
of course it didn't solve the problem.
So know I'm not sure what to do next.
Any other ideas?

Andy J =)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ardour.org/pipermail/ardour-users-ardour.org/attachments/20060503/882f58b8/attachment.htm>