I just searched for this in the forums, and there are many old requests for this. It’s nearly 2022 and the audio interface market has changed dramatically over the past ten years, but this is still an outstanding request.
I’m posting this in both the Cubase and ASIO forums as there are two different ways for Steinberg to solve this - inside the ASIO framework, or enhance Cubase to be able to talk to multiple ASIO devices at once. This post is to request the multi-ASIO feature inside Cubase. (edit - an admin deleted my post in ASIO so I guess this is where it lands now)
Specifically, I have a Universal Audio Apollo x8, an Eventide H9000R, and an SSL 2 connected into my Windows computer. I can use any one of these devices flawlessly through their ASIO drivers into Cubase 11 Pro, but never at the same time. It would be great to be able to bring a signal in through the Apollo X8 then send it to the H9000R for processing, then maybe back out through the SSL 2.
I have patch bays and a Heritage Baby RAM to handle a lot of flexible analog routing, but it just seems that I should have a lot more control inside the computer over what’s happening to my digital signal path.
ASIO revolutionized pro/prosumer computer audio when it was introduced, I’d like to suggest that enhancing it to support multiple audio interfaces is long overdue.
Two notes:
ASIO4ALL is not an option here as it uses WDM drivers under the covers and this provides neither the performance nor input/output count of an ASIO driver. I tried this and uninstalled it quickly when I experienced these limitations first-hand.
“Just buy a Mac” isn’t a valid solution either. One wouldn’t think that Mac migrations would be in Steinberg’s best interest either given the dominance of Logic Audio on that platform.
I just upgraded my license to 12 Pro this morning, lots of excellent additions.
I was hoping that this feature request would have made it in to 12 as it is quite a massive feature for those of us wanting a modern Pro-level studio setup.
This is not at all simple and, so far as I can tell, might be impossible to get working in the general case.
The biggest problem when using multiple DACs and/or ADCs is that there needs to be a single master clock. That is something that has to be done in hardware, either by connecting the clock from one interface to all the others, or connecting an external master clock to all the interfaces. Higher-end interfaces have dedicated clock I/O and/or the ability to receive a clock over a digital audio input. For example, the RME interface I have at this desk has both features, whilst the Sound Devices MixPre 10 II in my location sound bag has clock I/O but no digital audio. However, I am not sure all interfaces have clock I/O.
Even if you refuse to support interfaces without clock I/O, there are also potential issues with latency, as different hardware has different latencies. Trying to get sample-accurate performance across heterogeneous interfaces is likely to be impossible without manual tuning.
If you need high port count and modularity, it is likely better to use Dante, MADI or a similar high channel count digital audio bus with the appropriate I/O expanders, or use a vendor-specific interface chaining facility, such as that offered by Universal Audio.
I definitely don’t think it’s simple, but AVID, MOTU, and UAD have all done it (for their own devices) so it’s not impossible in Windows. As mentioned above, Apple has done it as well in OSX.
The best place for this is in the OS, perhaps Steinberg could put some pressure on them to see what they could do?
I have been looking into Dante more, and I just found that their Via software does EXACTLY what I’m requesting here. It aggregates audio devices and presents them as a single ASIO host to Cubase.
I’m assuming that if a small Australian company can write this software then it’s definitely possible for the highly talented developers at Steinberg to add this functionality into Cubase.
Using multiple hardware interfaces simultaneously and without a hardware clock sync is possible if sample rate conversion (re-sampling) is done in software to account for the clock drift. This would allow the use case described in the original post, for example. Of course, Cubase doesn’t support this now, but it could in the future.
ASIO4ALL aggregates WME (native Windows) drivers into one ASIO driver. WME is limited to only two channels, so if you have a multi-channel audio interface you are limited to only the first two channels.
Software resampling will lead to both wow/flutter in the input, and additional latency.
If you really need lots of inputs, then get something that’s actually built to do that well, like the RME devices. All your sound channels need to be on the same clock, and anything else isn’t really suitable for professional use. (Having written drivers for sound cards in a past life, I know this all too well!)
The degree to which there will be audible artifacts or distortion in the audio will depend on the quality of the SRC and the amount of clock drift. The added latency should be constrained to one additional DMA buffer. (As someone who has written drivers and firmware for audio interfaces, I know this too well.) Whether it is acceptable depends also on the use case. SRC is successfully used in many cases where HW clock sync is not available, including internally in some wireless HW designs and in Apple’s CoreAudio, which was cited in the original post.
I understand that this can be done in software when no other solution is available! Those solutions are always worse than doing it right in hardware.
My comment is more along the lines of: Why should Yamaha/Steinberg, who are in the business of supporting high-end production professionals, decide to spend their engineering resources on such solutions, when their target customers already have working hardware, and third party software solutions exist?
If you are a Steinberg customer, and are unemployed or otherwise unable to pay a couple of hundred euro/dollars/pounds for a multi-channel USB interface, yet require channel bonding between interfaces you have, then use a software solution from some third party. I think Steinberg/Yamaha should not spend any time on that themselves, compared to everything else that needs improvement in Cubase.
When Apple, Dante, Universal Audio and others have all done it in software why do you assume that is worse? As is mentioned above, if all the sources are properly clocked this will work just fine.
We aren’t privy to the user demographics, but Cubase doesn’t target only professional studios. The closest to what you describe is Avid, but even then they have targeted home studio users as well for what we can only assume is the economics.
Your assumption that my request is for “the unemployed or otherwise unable to pay a couple of hundred” is misplaced, I’m trying to get many high-end items to work together in my DAW and cost isn’t an issue for me.
What you’re proposing is a single-vendor hardware solution like MOTU (who also can be added to those that do it right in software, and also build it in to their Digital Performer offering). Your stated opinions exclude anyone from using more than one audio hardware vendor.
One of the options I considered was to buy two Apollo x16 interfaces, I’d have more than enough I/O and plenty of plugin processing too. However, I wanted to expand my palette to include Eventide effects, which you don’t think is something that anyone should want to do.
You are effectively saying that my use case is invalid, but I beg to differ.
jwatte, of course everyone is entitled to their opinion; this includes the original poster who put forward a solid case as to why no other solutions (3rd party or otherwise) are satisfactory to them. A point was brought up that what was asked for is not possible. I’m explaining that it is possible. Cubase could offer access to multiple interfaces simultaneously and only do software SRC when there is no shared HW clock (and SRC would work perfectly fine in many, many cases.) It’s a valid feature request from the original poster, IMHO. And it’s entirely possible, whether it’s something you personally care about or not.
In my opinion, what actual professionals (people who actually get paid by clients) need from tools like Cubase is what will allow them to deliver what the client wants. In my experience, clients generally care mostly about what they can hear (and what their clients can hear), how long it took to deliver, and how much it cost. So if this feature helps someone deliver that to their clients, I’m all for it.
Lguapo, I agree with what you are saying here but would respectfully point out that the market for Cubase is not purely professional studios and NAMM research shows that us “prosumers” make up a significant portion of “pro audio” revenues.
I’m not sure where it came from originally, but someone once told me, “Feed the rich, eat with the poor. Feed the poor, eat with the rich.” Please note this is a simplistic way to view this aspect of marketing and I am not advocating income inequality in any way.
Agree. In the end, Steinberg likely decides whether or not to implement this based on many factors. I’m just saying the arguments put forth that (a) it’s not possible and (b) it’s not useful to professionals are both invalid. My guess, too, is that Steinberg actually makes more of a profit from non-professionals than professionals (it’s just a much, much bigger market).
I think we disagree on the definition of “possible.” I include things like “work as well as the real solution” and “is a good business idea” into the overall. But if enough people would make a Cubase puchase with the feature, when they won’t, without, then the market can prove me wrong!
Recently I’ve seen a trend of people posting in various forums who are just contributing negativity and little else. This thread seems to have been infected by that. Why do people make the effort to communicate solely to attempt to invalidate others’ opinions? It really makes me wonder.