Windows computers, like Macintosh computers, require you pick one device as your principal audio interface. The principal audio interface is the one actually doing the A/D and D/A conversion to/from the computer... the one that connects to the speakers.

With Macintosh computers, you can build what is called an Aggregate Audio Device. This is where you can build a virtual audio object that includes the inputs of multiple audio devices. So, for example, the MONTAGE has 32 buses it delivers to the computer, the AG06 has 6 buses, you could create an Aggregate device that included all 38 Ins to the computer... the MONTAGE as audio interface provides 32-in/6-out, the AGG06 provides 6-in/2-out.... it is the stereo OUT bus from the computer (D-to-A) that is used to listen to yourself that is the issue ... typically a stereo pair you connect to your monitor speakers.

Simple right, the audio interface is responsible for taking in analog audio, turn it into digital signal for the computer, and it takes digital signal in from the computer and turns it into analog signal for your monitor speakers. So even with a Mac, that allows you to combine audio input, you will still need to designate one of the device as the device doing the D/A for the purpose of monitoring (hearing yourself). Only one of them can connect to your speakers

So no, the fact that the AG06 and the MONTAGE use the same driver, the driver “knows” the AG06 and “knows” the MONTAGE, when you open the Control Panel for the Yamaha Steinberg USB Driver all the devices that are using the driver will be identified on the Control Panel... on your Windows computer you will have to pick one or the other to do A-to-D and D-to-A, and even with a Mac (where as mentioned you could have 32+6 INPUTs) as your combined “aggregate” input device, you still have to designate one or the other to be the primary audio device (the one that connects to your speakers for L/R analog audio Out).

With Windows here are your options:
Since one or the other has to be selected as the audio interface, you want to setup a situation where all audio input sources in your studio connect to the audio interface.

If you select the AG06 as your principal audio interface, connect its main L/R Outputs to your monitor speakers, and connect all sound sources to its inputs. The Montage would connect (analog) to two of the six AG06 Inputs... and arrive in your DAW as a stereo source.

If you select the MONTAGE as your principal audio interface, connect its main L/R Outputs to your monitor speakers, and connect all sound sources to its A/D Input. The AG06 could connect its L/R out to the MONTAGE A/D Input... where you can route it as a stereo (or mono) Source in your DAW.

How to make the decision?
What do you need to record simultaneously? You could switch between a setup where MONTAGE was audio interface for laying down music tracks... then switch the cables that go to your speakers so they originate at the AG06 when you want to overdub items going through it.

I presume you are using the +48v phantom power supplied for condenser mics... but usually, you are overdubbing when using microphones. Decide if swapping cables is feasible. Everyone’s studio setup is different. For some this is not a big issue, for others it simply is not possible... but the thing to understand, either device can playback audio created by the other (after years of explaining this stuff, I’ve realized this needs to be stated - so forgive me if this sounds like a “well, of course”... if you record first through the MONTAGE as audio interface... and then switch to the AG06, the audio you recorded through MONTAGE will play just fine through whatever audio interface you choose. And vice versa.

Once the job of D-to-A is accomplished, any audio interface should be able to play it back... so swapping is a matter of changing who’s doing the conversions.

If using the massive potential of the 32 audio buses out from the MONTAGE appeals to you, recognize that in a typical multi track workflow, you are laying down music tracks first, then adding other instruments as overdubs. Only if you need to record everything or several things at once would you need to hook both the MONTAGE and the AG06 up simultaneously. So that’s the question you must ask yourself:
Do I need to record signals from both the MONTAGE as a source and the AG06 as a source simultaneously?
Is it logistically feasible to swap audio interfaces? Some DAWs, like Cubase, allow you to do this without even powering down the program.

Thank you for the detailed answer.

Two more questions please :

Is it expected that the Steinberg Driver soon offers the same "Aggregate Audio Device" under Windows (only for the Yamaha products, of course)?

Do we know the round-trip latency of the Montage audio interface? Is there a Montage test analyzing latency and giving the best buffer size to avoid bad recordings? I consider to buy a montage and use it as main audio interface, but I'm still undecided.

Thanks.

Is it expected that the Steinberg Driver soon offers the same "Aggregate Audio Device" under Windows (only for the Yamaha products, of course)?

There is no such thing as a “Steinberg Driver”, we’re sure you mean Yamaha Steinberg USB Driver... your question is one for MicroSoft as it is their operating system that does not allow multiple audio devices. It is Apple’s operating system that does allow multiple audio devices...

In short, that’s a Windows thing.
So you need to record simultaneously through two different interfaces?

Do we know the round-trip latency of the Montage audio interface? Is there a Montage test analyzing latency and giving the best buffer size to avoid bad recordings? I consider to buy a montage and use it as main audio interface, but I'm still undecided

The latency is really a function of your computer. There is no need for a Montage test, your DAW will report on the Input and Output latency according to the Buffer Size you set. “Bad recordings” vs.” good recording” is going to decided by your ears, and is how to best set the Buffer Size based on your computers ability to receive and put back out the audio. Cubase AI (the free entry level DAW, comes with the MONTAGE) will analyze the input and output latency out to three decimal places. When recording audio with MONTAGE, latency is a non-issue because you can monitor yourself direct. You never have to be at the wrong end of audio latency while recording MONTAGE.

No doubt latency comes with the computer and the buffer size, but a little bit with the Montage too. The question was about the round-trip latency between the A/D input of the Montage to the PC/DAW (without any VST, just the DAW, Cubase for example) and return to the main D/A output of the Montage, without direct monitoring. No information about that ? This is important if the Montage become the main audio interface of the setup.

From Cubase you can open the Yamaha Steinberg USB Driver properties and view what the calculated latency is for your current buffer settings.

Cubase AI (the free entry level DAW, comes with the MONTAGE) will analyze the input and output latency out to three decimal places.

This shows a different driver (and older version of Cubase) - but the basic procedure is the same as far as viewing calculated latency.

https://www.youtube.com/watch?v=9LxZ7dIb7Yg

I didn't watch the whole thing - just the part where he clicks on the menu option to view the driver properties (and the prior menu which also has calculated latency).

Here we have the latency when using the Montage Audio Interface and the Steinberg USB driver on a Mac:

https://youtu.be/vvT6zXOb5W0 (4:27)

The round-trip latency is about 6.7ms without counting the buffer size (256/44.1). It's not bad but not great either and could be better.

When recording audio with MONTAGE, latency is a non-issue because you can monitor yourself direct. You never have to be at the wrong end of audio latency while recording MONTAGE.

You are right. But if you can't work with direct monitoring (because you have to use a VST that is not built in the Montage), latency is obviously a problem.

The Steinberg USB driver could and should be optimized, like the RME drivers for instance. If it's possible for RME, it should be possible for Yamaha. Steinberg is the creator of ASIO. It is unfortunate that it does not offer the best equipment today. It is unfortunate that multiple audio interfaces are not supported by ASIO driver (not a problem of Windows, but only with ASIO).

It's a problem that Yamaha does not communicate about driver latency, or A / D and D / A audio specs. Potential customers who pay a lot of money for the Montage expect detailed, clear and objective datas about every part of the Montage.

If the hardware and software was less general-purpose, then published latency specs would be more meaningful. Right now you can only say:

VSTi latency=??? (depends, some add a lot)
+
Computer hardware/software/OS latency=??? (depends on what else you have loaded, amount of swapping, CPU type, hops between CPU and USB bridge, BIOS settings for low-level chipset/USB, driver buffer settings, etc)
+
Yamaha Montage USB/ADC(or DAC depending on direction) path through to tone generator = ### (this could be defined - probably a number you're looking for)

Once you have some number defined for Montage's contribution, that's a small part of the pie. The rest you'll be tasked with figuring out on your own.

This isn't preventing Yamaha from buying a couple motherboards (for PC) or a couple of different Macs with published CPU/memory configurations, loading the latest operating system, only installing Cubase + Drivers, and publishing the latency numbers for that configuration. That'd let you know at least if you had the same computer (Mac) or motherboard (PC) then you can expect something close to the published latency numbers. Also publishing any settings to help accomplish this like buffer size.

Often when this is done, the manufacturer will find "hacks" to get the numbers better - registry hacks, etc. So it's not a perfect world even if published since some of the settings to get a low latency may "hurt" performance of something else you need the computer to do.

I can understand since so much is out of the control of Yamaha (computer hardware/software config) why the numbers are not currently published. Even if numbers were published -- it's a never-ending cycle since hardware and software changes almost on a daily basis (new OS updates, new hardware CPUs and chipsets).