Re: [Jack-Devel] is this latency coming from ALSA, jack, or ardour?
Le 14 avr. 2011 à 05:57, Chris Caudle a écrit :
> On Wed, April 13, 2011 1:03 pm, Chris Caudle wrote:
>> My hunch at the moment is ALSA, but that is why I
>> am experimenting, to determine that empirically.
>
> I think I have confirmed that the latency reported by ALSA to JACK is the
> problem, but I would be interested to hear if anyone spots something I've
> overlooked.
>
> So my previous session bouncing from track to track analog out back to
> analog in with my USB interface ended up with each track offset from the
> previous by about 13ms.
>
> Using my PCI Envy24 based card (M-Audio Audiophile2496) each track was
> offset from the previous by some small fraction of a ms. The total offset
> from track 1 to track 10 was in the range of 0.5ms to 0.7ms. Using a
> recording of the click track, I could just hear slight doubling between
> any pair of tracks. On any more realistic scenario (i.e. not comparing
> identical tracks to each other) I think it would probably be so close it
> wouldn't matter, there would be more variation in a player's rhythm than
> the offset, so you shouldn't be able to tell.
>
> I think what that says is that the M-Audio PCI card is reporting a latency
> figure back to JACK that is within some fraction of a millisecond to being
> exact, and the USB interface is reporting back some bogus number that is
> off by over 10ms. That, or there is some additional latency in the USB
> stack that the interface can't know about, and ALSA is not compensating
> correctly and adjusting the number when it passes the information to JACK.
> I don't know how JACK gets the latency information for the analog
> interfaces, so I'm not sure where the inaccuracy could come in for the USB
> interface. I think the short of it is I won't be using my USB interface
> to overdub percussion, unless I'm prepared to do some manual shifting to
> get everything lined back up.
>
> It also brings up a question of digital interfaces. If I use the digital
> out from the card with external ADC and DAC, that is additional latency
> that the card has no way to know about. Is there anyway to tell JACK (or
> Ardour) about the additional latency so that it can be automatically
> compensated like it is when the analog interfaces of the card are used?
>
> --
> Chris Caudle
>
>
With a running JACK server, can you try:
jack_lsp -A -l
and report what is printed?
Thanks.
Stéphane
1302760168.314_0.ltw:2,a <1F99B6B7-F32C-449A-A9B3-551F14CEEAE8 at grame dot fr>