I wanted to share a simple process that I’ve been experimenting with, continuing the discussion from here:
So, reamping is a no brainer when you have multiple outputs in your interface, but that is not my case, and researching about using two interfaces within Ardour taught me a little bit about clocks and the mostly certain desyncing issues I would encounter that way. Now with Ardour 6 I know it is possible to use different devices for playback and recording although don’t really know how much safer that is, my guess is it’s still kind of risky.
Well my approach is just to use two different programs, each one controlling one interface. So I had my guitar DI recorded and edited, then exported to a file that I would play on Audacity from interface 1, that signal would run through a reamp box, then into the amp, which will be recorded through a mic connected to interface 2, controlled by Ardour. This gives me a recorded signal without actually needing Ardour to handle two interfaces at the same time, and I would think that solves the syncing problem for me.
So far it is working great, and spares me the need of buying another device when I already have two cheaper ones laying around. Anyway I may be missing something so I look forward to the more technical people’s thoughts.
No, Ardour is using a resampling library to keep the audio synchronized on the file side. In essence it is running a sample rate converter to adjust to the small difference in clock rate between the two interfaces.
That will result in the recordings eventually being out of sync. If you are only doing that for a couple of minutes at a time the difference in file length will probably not be noticeable, but it is definitely doing things the hard way compared to using the built in resampling in Ardour 6. I would consider that the hard way even compared to setting up jack and zita-a2j resampling, but certainly with the capability built in to Ardour6 now there is no reason to consider using multiple applications like you are, just do everything in Ardour.
Cool, thank you for your answer, I am definitively gonna try everything inside Ardour. Although I cannot understand why could this be out of sync, I am clearly out of my thing here but I thought the problem was with one program handling different clocks at the same time, this on the contrary is to me like Ardour handling its thing and then being fed something external with no sync conlifcts. What you say is then that the problem lies in the computer itself processing audio with two devices?
Each audio device has an electronic clock to generate the signals required for sampling the audio to digital values. A typical sample rate is 44100 times per second used for compact disc, meaning that the audio signal is sampled once every 22.7 microseconds. The two interfaces would have to sample the audio synchronized to within a fraction of that 22.7 microseconds, not possible with standard electronic clock devices without making an electrical connection between the two devices to keep them synchronized (for example some interfaces have a digital audio connection which can be used to synchronize one interface to the other).
I thank your disposition to explain these things, eventhough I’m still a bit confused. What you say is clear to me in a context of recording a track while hearing a backing track, everything must be synced to a ridiculously small fraction of time for it all to work fine, because both things have to be later mixed and played at the same time. Even in the context of reamping, when one program is handling two interfaces we have to make sure those two are talking to each other in sync, because one is directly recording what the other is playing inside the same software, and should they not be in sync what has been recorded may not correspond to what was played to the sample. This is what I understand Ardour is now capable of doing by itself thanks to latency compensation improvements.
Coming back to “my solution”, I would like to understand how two different softwares doing two different things each one on its own, with two different devices can have a sync problem, even more I don’t see why they have to be in sync as they are indeed working on two completely different things that don’t have to talk to each other. Let’s say I take a computer just playing on speakers something I would then record with another computer, the two systems are no way synced, but is there a sync problem if I go on and put what I recorded on top of what I played and mix it together?
In the case of something short like a typical pop song, likely there will not be a problem. The case where it can cause a problem is if you play a long file through one device, record something new through a second device, and then when you import both the original and the newly recorded file you find that the lengths are very slightly different. And I do mean very slightly, for typical clock components running at normal room temperatures (not at extremely cold or hot conditions), the clocks between two decently built devices should be within 0.001% of each other. Over the course of 2 minutes that is 1.2ms difference. Over the course of an hour that is 36ms difference. So if you were recording some long instrumental piece in a single hour long take there would be noticeable offset by the end.
For a couple minute pop song, no problem (other than it is somewhat cumbersome using two different applications for overdubs like that).
