gimmeapill wrote:
The resampling however does happen, since you go out of the Guitarix soundcard to the qtracktor soundcard using analog jacks.
So if I understand the signal chain correctly, the signal of the guitar goes
Code: Select all
Guitar->AD->Guitarix->DA->audio cables->AD->Qtractor-> DA -> main out.
|-> Direct monitoring
Yes, resampling does happen but not in the PC (alsa_in/out, audioadapter ...) what brought lots of problems when I tried.
gimmeapill wrote:
But if this really allows to isolate the DSP loads and run several applications at different latencies on the same system, then I think you just invented "audio workload containerization"
The requirement for dedicated audio interfaces of different models is however annoying.
So I'm still wondering if it wouldn't be possible to achieve the same result yet only in software by having the low latency jack instance using the physical interface, and the high latency one using a virtual loopback as a dummy interface and zita-bridges connected to the first instance.
I don't know why there is no way to get the Scarletts' serial numbers to distinguish between them. It's annoying.
Dire Straits sing "Check out Guitar-George, he knows all the chords".
I sing "Check out bluebell, he tried all the resamplings (and got all the problems you can think of and even more)"
Alsa_out crashed as soon as I started Audacity (which seems to kill alsa_out when scanning for interfaces).
Similar thing with zita-stuff.
Loading audioadapter into jackd kept increasing the ringbuffer (and so the latency) without any observable reason.
Using a network connection on the loopback interface to couple 2 jackds, bad latency.
But I made these experiences when trying to use 2 audio interfaces that had no means for hardware sync (Scarlett 2i2, 2i4, Solo). Maybe you get better results with your attempt with only one audio interface.