[Ardour-Users] Hydrogen/Ardour sync problem

[Ben Saylor]

Hi,

On Monday 28 January 2008 05:39, Paul Davis wrote:
...
> > > I've also encountered this problem.
> > > The time-shift depends on the frames/period and periods/buffer
> > > settings in jack. It's been a while since I investigated this, but as
> > > far as I remember that's what I found.
> >
> > Which means its constant (in your session of jack) and simply because
> > of latency compensation happening / not happening. Nothing to really
> > worry about, ardour just can't tell wether the sound its
> > playing/monitoring is from inside or outside the computer...
>
> actually it can, and it does. ardour will trace the signal path back to
> a physical port (if one exists).
>
> however, the version of JACK is important. although the basic
> information required for latency compensation has always been available,
> there are some little details about how and when its computed that
> affect the results.
>
> what version of JACK is being used in the scenario where this is not
> working?

I have this problem as well.  I just did a small experiment with Ardour 
2.0.3, Hydrogen 0.9.3, and Jack 0.103.0.  I set Ardour as the Jack 
time-master and turned on transport sync in hydrogen.  I set all humanize 
controls to 0.  I set Jack's buffers to 256*3.  After recording a few bars 
and looking at the waveforms in Ardour, it was clear that the drum track 
was recorded slightly ahead of the beat (i.e. the attacks were offset 
slightly to the left of where they should be).  I tried with Jack at 
2048*3, and the the degree of offset was greater.  At 256*24 (same total 
buffer size as 2048*3), it had exactly the same offset as with 256*3.

Ben