Multitrack Recording?

I’ve been using Ardour for simple mixdown and voiceover production work (and love it!), but how to use it’s multitrack capabilities to record multiple tracks in a session? The largest interface I’ve seen is 8x8 mic channels, although there are some with added line level instrument inputs. Can one use multiple USB interfaces simultaneously, or are there dedicated multitrack PCIe cards for this?
Thanks in advance,
Ardour5, Ubuntu Studio 19.04

One thing you can do is get something like the Focusrite Scarlett 18i20 along with the Focusrite Scarlett OctoPre, which acts as an expansion. Each one has 8 inputs, giving you a total of 16 inputs. As far as Ardour is concerned, that setup would be treated as a single interface. I’m certain that there are other possibilities out there.

1 Like

There are dedicated interfaces for this, MADI for example has 64 channels I/O. It doesn’t have to be PCI[e], you can also get USB or Thunderbolt interfaces: for example

There’s also a discussion on: Linux AVB Support

It is not impossible to use multiple USB soundcards, however this is not recommended, especially not for recording. All those devices have different internal clocks. Say one soundcard samples at 48000.001 Hz and the other at 47999.997 Hz. Over time they drift apart. The audio needs to be re-sampled in order to synchronize them, which affects quality. Depending on your use-case, this may or may not be acceptable.

For this reason, most high-end processional soundcards have word-clock I/O to synchronize the soundcards in hardware to allow chaining them or using an external clock-source. However this is rarely found on USB devices, since USB itself also complicates synchronization.

A longer explanation can be found at

Anyway, how many channels do you need?

For a home-studio, perhaps follow @Michael_Willis advice and investigate Focusrite 18i20 or look into PreSonus Audiobox. In some cases an ADAT breakout box can help to get a few more channels, too.

The Behringer UMC1820 is also fully compatible with Linux. I have an ADA8200 connected to it via ADAT that gives me a total of 16 inputs. The Focusrite 18i20 is no doubt a higher quality device, but getting its internal mixer to work on Linux can take some effort from what I’ve read.

1 Like

I use exactly the equipment that @Michael_Willis specified and I have ZERO problems on linux (Ubuntu Studio uh… 16.04?). The second generation Scarlett’s (and the third generation that got announced today) may be a different story. But first gen 18i20 and the Octopre MkII will give you no problems and are really great soundcards for the money.
As @x42 asked, how many inputs do you need? What kind of multitrack recording are you getting into?

Yes, the first generation 18i20s have ALSA support, so they are fully compatible. Work has been done to get the second generation’s mixer functional, but I don’t think it has been integrated into the kernel yet, so some extra effort is required (i.e. fetch some code from GitHub). I didn’t realize a third generation had been released. Time will tell what its story is, but the manufacturer’s page makes no mention of iOS compatibility, so Linux users should proceed with caution.

Thanks everybody!

It was a general question, I was thinking of small choirs and Church work, although I don’t have a definite application yet. I got burned recently dropping a basketfull of cash on a GPU that turned out not to work with Linux…

16 inputs would probably do the trick. Or record on a Tascam 24 and mixdown in Ardour. Currently using a Behringer Firewire 610 in the studio, which works fine but for the occasional xrun.

I suggest to use a stereo microphone setup for this, and maybe a spot mic for soloists.

That is assuming you also record in the chuch (not a church-choir in a studio).
With multiple mics you would have to close-mic all individual performers - otherwise you’ll capture the space (reverb) at different points in space/time, and also introduce comb-filter effects on neighboring mics capturing the same vocalists.

You probably want to just use 4 inputs, and invest in good microphones (and mic preamps) instead.

1 Like

One solution not mentioned yet is some mixers have USB connections such that they act like a large audio interface in addition to the usual audio mixing tasks they perform. If you need a mixer for other reasons, it is a small relative increase in price to get one with a computer interface.

To use multiple soundcard you need to use alsa_in/alsa_out. If they can share a wordclock it is even better.
So if it needs to done you can use that also the jack discussion on multichannel is outdated type man 1 alsa_in.

Do you know where I can find the mixer work for the 2nd generation? Also with the 18i20 I cannot figure out how to get the 18i20(2nd gen) to work past 48kHz. The adat gets disabled so I know all the inputs do not function properly but, I get no soundout of any. You have any ideas? Does this not happen on the first generation?

As mentioned above, I’d advice against this for the case at hand. Not only will the input be resampled, but even if you use word-clock sync, there will be different systemic latencies.

This can be fine for some rock/pop band production where you don’t record the same source, but capture different musicians and different instruments where the signals are not correlated.

However, if you capture the same space with different microphones (here: choir in the church) it will introduce artifacts. Depending on used mic-techniques you can get a skewed stereo-image or comb-filter artifacts.

PS. you may prefer zita-ajbridge (or jack1’s built in zalsa). compared to alsa_in/out it offers both higher quality and requires less CPU power. zita-ajbridge does also allow to configure systemic latencies.

1 Like

The first and second generation Scarlett 18i20’s ADAT input supports 8 channels in up to 48kHz and only 4 channels in 96kHz and above. It is the bandwidth limitation of an ADAT cable. To make matters worse, where the 4 ADAT channels get mapped is seemingly random. There were times where my Octopre’s 3rd input would be channel 13 in Ardour and times when it would be channel 15 or 16. It was a mess. I ended up switching to 48kHz sample rates for all my projects.
Now, if you’re okay with adding no more than 4 input channels via ADAT, you should be able to get the Scarlett to sample at whatever rate. I’ve recorded plenty at 96kHz both with and without the Octopre. When you start your project, are you selecting 96 or 192 kHz as your sample rate? I personally use the first gen, but my friend has the 2nd gen 6i6, and I believe he is recording at 192kHz without an issue.

I have to used aggregate device for testing at work and have been using alsa_in but, I completely forgot about zita-a2jbridge thank you so much for reminding me about it. Also I ran a test with my m-audio delta 1010lt and Saffire 18i20 the clock was stable when I used spdif as clock source for the delta 1010lt. I did not check the phase of the two device inputs but, yeah it could easy introduce a variably combing. I see the jitter in estimation with the printout from alsa_in. I am now going to mess with zita-a2jbridge at work now.

So I have a github but, where is located can you send me a link?

The link is in my post above, just click on “geoffreybennett/scarlett-gen2”. There are rpm and deb packages on that page under ‘Assets’, as well as the source code. I don’t own this device, so I have never installed this custom kernel, but it looks like you would install it like any other kernel then create the .conf file as noted in the instructions on that same page.