Using Aggregate Devices to chain input interfaces - does it work in AA?

Report · Oct 18, 2022

I currently record with a Focusrite Saffire 2i2 interface that allows for two mic inputs, with a USB-C output. Works like a charm with AA...but now I need a third mic. Uh oh.

Unfortunately Focusrite's next step up, the 4i4, still only has only two mic channels (the added ones are INST only). So I am looking into binding two of the 2i2's together using the Mac OS Aggregate Device capability. See https://support.apple.com/en-us/HT202000 for a succinct How-To from Apple.

On the face of it, it sounds like exactly what I need: although each device is coming in through its own USB-C port, the CoreAudio subsystem should treat it as a single 4-input device, with AA none the wiser. But before I order a second unit, has anyone actually tried this? Caveats or warnings??

Report · Oct 18, 2022

My only comment on this relates to all aggregated systems, really, if they are USB- linked. The issue is that technically the two devices you are aggregating aren't actually going to be sync-locked. What the aggregation process does, amongst other things, is uses one device as a master sync source and 'corrects' the timing from the other one digitally so that it appears to be running at the same rate and in sync. Normally this isn't a problem, but if, for instance, you tried to make an acoustic stereo recording with one mic on one device and the other mic on the other device, you'd get phase shifts - almost certainly audible after a while. This is essentially what point 6 in the How-to is about. If you don't enable it, what will happen when the clocks drift one step apart is that you'll get a very loud 'click' in your recording. How often these clicks occur depends entirely on how close the two clocks are to running at exactly the same frequency. It's never perfect, and the spec for clock accuracy is surprisingly lax with a lot of devices, especially at the cheap end of the sound device spectrum.

If, on the other hand, you're just putting three mics up in a recording space and have three people talking and the drift correction enabled, the chances are that you won't notice a thing and it will all be fine.

Report · Oct 18, 2022

This is helpful, thanks, Steve. Would you expect an issue with three mikes recording a single audio source (guitar)? Same Focusrite model, same generation? Not asking for a promise ;)...just your best judgment.

Report · Oct 18, 2022

Really it depends upon how closely matched the two clocks are. Unfortunately, just because they are identical devices, it doesn't mean that the clocks will be - that's all down to the individual crystals in them. But if you are going to have trouble, it's definitely going to be from single sources, because (obviously) the source sound is in sync!

But if you want a clue, consider that sound travels at approximately 1ms/ft in air. This means that for every foot further away you put the mic, (at 44.1k) another 44 samples delay gets added anyway as a 'timing discrepancy'. That's actually quite a lot, compared to the sorts of errors you'd commonly get, so almost certainly it isn't going to be a problem, I'd say. More likely to be an issue with a large scale stereo recording where there's a lot of room sound. The other issue with the clocks is about their stability. If they are both relatively stable, even though there's a difference in the actual rates, then any phase issues will be far less likely to be noticed anyway; they generally manifest themselves when the changes in them happen fast - which would be far less likely in this case.

Report · Oct 18, 2022

Thanks for taking time to educate me. That all makes total sense. I think I'll give it a go and see what happens!

Report · Oct 18, 2022

I should add that I did successfully test this using my Mac's microphone as a third source. AA recognizes the aggregate device and shows three inputs. (See screen shot.) I just want to verify that it will work with another Scarlett connected through the second USB port. Don't see why it wouldn't if CoreAudio "accepts" the existence of that virtual device, but wanted to see if anyone had actually pulled it off IRL. Thanks - Edward