Using multiple audio interfaces

[benjaminsbowers]

I'm using my Axefx and line6 ux2 to record, but I don't want to have to switch between devices in Reaper to listen on my monitors after tracking with headphones on the active device. For example when im tracking someone else and determining a good take, I have to switch devices in preferences or use their headphones to listen after the take, rather than listening while it's being tracked. Would I just use a mixer and go LR out of my devices into the mixer, then main LR out of the mixer to my monitors? Or is there an easier solution? My devices are connected to my computer by USB alone currently, and my monitors are plugged into my computers speaker input. I apologize in advance if there is some obvious solution I'm not seeing here. Hope this made sense haha.

[Fabian]

Originally Posted by benjaminsbowers

[...] I apologize in advance if there is some obvious solution I'm not seeing here. Hope this made sense haha.

Don't apologize, this is a question that contains some controversy... Some people will tell you that it is not possible, because ASIO does not allow several drivers simultaneously. However, try ASIO4ALL (just google), it is a single driver that can drive several devices. I have used it to drive three different devices simultaneously, and it works. It may not be able to give you the lowest latency, but on my machine it is good enough (in fact, it is better than the native drivers for my audio i/f).

[4x4uk]

This plus one, I use 4 separate audio devices together in reaper all the time

[benjaminsbowers]

Originally Posted by Fabian

Don't apologize, this is a question that contains some controversy... Some people will tell you that it is not possible, because ASIO does not allow several drivers simultaneously. However, try ASIO4ALL (just google), it is a single driver that can drive several devices. I have used it to drive three different devices simultaneously, and it works. It may not be able to give you the lowest latency, but on my machine it is good enough (in fact, it is better than the native drivers for my audio i/f).

Cool, thanks for the tip. I'll check it out!

[serr]

Make what is called an aggregate device with your OS utility app (Audio MIDI Setup for OSX and ASIO4ALL for Windows). This combines your multiple interfaces into a single virtual device. Select this aggregate device in Reaper (Preferences/Audio/Device page) instead of one of the single interfaces. You now have access to all I/O from all devices in the aggregate.

Works 99.9% of the time flawlessly in OSX. Responses from Windows users indicate it might be more involved to get working.

In either OS you will have to pay attention to the clock/sync connections and settings of the hardware devices.

[LloydPDX]

Originally Posted by serr

Make what is called an aggregate device with your OS utility app (Audio MIDI Setup for OSX and ASIO4ALL for Windows). This combines your multiple interfaces into a single virtual device. Select this aggregate device in Reaper (Preferences/Audio/Device page) instead of one of the single interfaces. You now have access to all I/O from all devices in the aggregate.

Works 99.9% of the time flawlessly in OSX. Responses from Windows users indicate it might be more involved to get working.

In either OS you will have to pay attention to the clock/sync connections and settings of the hardware devices.

@serr, thanks, just what I was looking for, even though it's a couple of months since the OP raised the question. Are you able to shed some more light on minding the sync and settings? I'm using an iMic interface with OSX 10.9.5 for tracking an old cassette, but haven't been able to get playback through my Scarlett 2i2. (Unless I unplug the iMic; then playback is fine.) These two devices have different sample rates and bit depths (the imic is 44.1/16), perhaps that's the issue? I'm still a bit mystified when it comes to setting up these aggregate devices and knowing how to get these things to line up. Do the project settings come into play here as well?
Thanks for any pointers you can offer.

[serr]

Originally Posted by LloydPDX

@serr, thanks, just what I was looking for, even though it's a couple of months since the OP raised the question. Are you able to shed some more light on minding the sync and settings? I'm using an iMic interface with OSX 10.9.5 for tracking an old cassette, but haven't been able to get playback through my Scarlett 2i2. (Unless I unplug the iMic; then playback is fine.) These two devices have different sample rates and bit depths (the imic is 44.1/16), perhaps that's the issue? I'm still a bit mystified when it comes to setting up these aggregate devices and knowing how to get these things to line up. Do the project settings come into play here as well?
Thanks for any pointers you can offer.

Digital 101:
You must run the system from one sample rate clock and one sample rate clock only. Otherwise things drift, the samples don't line up, and you hear mad clicking, popping, and distortion. ie. It straight up doesn't work.

You can connect 49 different audio interfaces and other digital devices but you must always pick one of them to use as clock master and slave all the others to it.


If you have a digital device (the iMic) that ONLY supports 44.1k sample rate, then that's your only choice for a system containing that device.

If you have a device in the system that has no ability to sync (slave) to another (because it doesn't have word clock input or a digital input to clock from), then you will have to make said device the clock master and sync everything else to it. Otherwise, choose the device with the highest quality sample rate clock to use for the master.



Reaper gives you literally every option for controlling your various devices from setting these parameters from Reaper to setting them from OS control panels (eg. Audio MIDI Setup or ASIO4ALL) to other audio apps or proprietary control panels written for some devices.

That iMic has no inputs of course. I imagine your only option would be:

Aggregate device of iMic & Scarlett 2i2 (check the box for the iMic 1st when creating).
Try controlling the aggregate device from Reaper first. (Check the boxes for sample rate and block size and enter the values. If the iMic only supports 44.1k, then 44.1k for the system it is.) Check the box and set the project sample rate the same (in Project Settings).

Everything playing nice?

If not, the 2nd choice would be to uncheck the boxes in Reaper preferences to allow control from other apps or control panels. Try Audio MIDI Setup to control everything. Again, check the box and set the project sample rate the same (in Reaper Project Settings).

[Tod]

Okay, I just got another small new interface today and I did manage to get both the old one and the new one setup to use together.

As mentioned I used ASIO4AL and at first I couldn't get it to work. But then I clicked on ASIO Configuration and managed to get both interfaces selected and then it did work. I did have to change the buffers from 252 to 512.

Incidentally, one of the interfaces is a PCI card and the other is USB.

It's only on rare occasions I'll need more inputs but when I do, it's nice to know I can get them.

[scottearth]

Super old thread here, but it came up on a search.

I have a macbook. I want to run 2 8 input interfaces. Is Mac the same process and windows?

[nait]

Originally Posted by scottearth

Super old thread here, but it came up on a search.

I have a macbook. I want to run 2 8 input interfaces. Is Mac the same process and windows?

I don't own a Mac, myself, but I remember this being discussed in the last year ish . funny enough, I think it was by Serr, who posted the same info here in this thread on post #5. I don't know anything about it, but I remember a post elsewhere about being able to set up an aggregate device on Mac. It stuck with me because it's one feature I was very jealous of!!! lol

So the short answer is it actually sounds simpler and perhaps more reliable on Mac.

[serr]

It's a two step process:

1. Make the aggregate device to combine the multiple interfaces into a single virtual interface.
2. Choose one of the interfaces as sample rate clock master. Slave the other interfaces in the aggregate to the master.

The 2nd step is what loses people I suspect.
Also if you have an interface with no word clock or digital I/O, you're left to try sync over the data connection (USB/firewire/TB/logic board integrated) and this may not give the best stability. Or it may only work at SD sample rates. etc etc

Multiple interfaces all with word clock and properly configured are rock solid stable. Using a digital audio connection to also carry sample rate clock when word clock isn't an option is also usually just as rock solid stable.

[EcBaPr]

I don't have ASIO4ALL installed on this computer to check but from what I recall it doesn't have any options to define one device as master clock and others as a slave in the driver does it ? so in order to set a master clock device wouldn't you need to physically connect BNC/ADAT cables between the different hardware ? and then set up the clock settings for each device in its own control panel ?

or can ASIO4ALL do sample accurate sync in software itself with just the driver ? if it cant then how would you sync multiple devices over just a data connection ?

[serr]

You need to physically connect the cables. The controls for an interface's sample rate clock connection are usually found in a proprietary control panel app delivered with the device. Or they may be front panel physical switches on the device. You would not expect to find any way to control these parameters with your OS audio control panel. (Audio MIDI Setup on Mac or ASIO on Windows) The OS audio control panel is only used to create the aggregate device config. On the Mac, you get a "resample" checkbox. This is the option you get stuck with if there's no word clock or digital I/O with the caveats as mentioned. I don't know what this part looks like in Windows.

[EcBaPr]

yeah connecting a sync cable and setting up each device as master/slave is what I have always understood to be needed for sample accurate sync.. I think the way ASIO4ALL is implemented potentially creates the impression that once your devices are aggregated they are in sync also.

[Tone Deft]

Originally Posted by EcBaPr

yeah connecting a sync cable and setting up each device as master/slave is what I have always understood to be needed for sample accurate sync.. I think the way ASIO4ALL is implemented potentially creates the impression that once your devices are aggregated they are in sync also.

nah, it's a lovely piece of software.

the ONLY way to see if you clocks are in sync is to make a copy of a track and see if they phase invert and cancel at the right volume level.

sample accuracy is a farce in home computing DAWs, not with the budgets we have.

ASIO4ALL is a great tool. with Ableton I've had it in a stuck state so I uninstall and reinstall, all's well.

edit - I gotta add that just to design your own audio products with multiple input sources requires an SRC or ASRC to get the word clocks lined up. switching between inputs can absolutely wreak havoc on your audio if you screw it up. you switch streams at the end of an audio word, bit block can sync up but Word clock ends up as a runt pulse (think 1% duty cycle) that can be caught by some devices but not others. it's well beyond the layer of hooking up boxes.

[drumphil]

I don't know what this part looks like in Windows.

Windows audio doesn't have a feature to resample devices with no hardware wordclock link to keep them synchronized. Neither does ASIO4ALL.

I'm curious to know how usable the mac audio system is at low latencies when using resampling to maintain synchronization, because resampling necessarily takes time, adding to latency.

I'm also curious to know how or if it accounts for the varying levels of input and output latency between different audio interfaces at the same buffer size. How well aligned do the tracks from different interfaces end up?

Is this feature usable for real time audio, or is it mostly just to allow cheap audio interfaces to be used simultaneously for applications where the above concerns aren't critical, like podcasting with a USB mic, using the mac's headphone out for monitoring?

sample accuracy is a farce in home computing DAWs, not with the budgets we have.

Isn't there a program that combines multiple ASIO drivers? Not like ASIO4ALL using WASAPI, but actually using the native ASIO driver for each interface. I think there is now, but I can't for the life of me remember what it's called.

Doing it like that, assuming that each device is accurately reporting it's input and output latency, it should be possible to achieve sample accuracy. But how many devices report accurately, and how many of them maintain consistent latency at a given buffer size. That bit I don't know.

[Tone Deft]

Originally Posted by drumphil

Windows audio doesn't have a feature to resample devices with no hardware wordclock link to keep them synchronized. Neither does ASIO4ALL.

I'm curious to know how usable the mac audio system is at low latencies when using resampling to maintain synchronization, because resampling necessarily takes time, adding to latency.

I'm also curious to know how or if it accounts for the varying levels of input and output latency between different audio interfaces at the same buffer size. How well aligned do the tracks from different interfaces end up?

Is this feature usable for real time audio, or is it mostly just to allow cheap audio interfaces to be used simultaneously for applications where the above concerns aren't critical, like podcasting with a USB mic, using the mac's headphone out for monitoring?




Isn't there a program that combines multiple ASIO drivers? Not like ASIO4ALL using WASAPI, but actually using the native ASIO driver for each interface. I think there is now, but I can't for the life of me remember what it's called.

Doing it like that, assuming that each device is accurately reporting it's input and output latency, it should be possible to achieve sample accuracy. But how many devices report accurately, and how many of them maintain consistent latency at a given buffer size. That bit I don't know.

hold up. you know your DAW is being sample accurate when you can send a signal out to a new track that's phase inverted. as the volume on either track goes up or down the overall output from the DAW should also go up and down.

ASIO4ALL is the standard app that can talk to multiple audio drivers at once. it's been around for years and no other software is more recommended.



loosely speaking, the off-topic nuts and bolts not meant to describe all this in detail, just the gist of what's going on.

at 48kHz sampling rate each sample is taken 1/48kHz = 20.2uS. that means that the analog audio gets sampled into digital every 20.2uS.

taking in data from various sources has two problems:
1 - crystals are electronic devices used to create highly accurate clocks. even then those have a tolerance. a 48kHz crystal itself has some tolerance to it. IOW no two 48kHz clocks are EVER truly in sync, it's a problem with electronics. so that 20.2uS is the expected rate but it's NEVER 20.200000000000000000000000000uS exactly.

2 - that means your system takes in data at different rates, that is handled by a device called a Sample Rate Converter (SRC). a higher quality, slower type of SRC is an ASRC (asynchronous sample rate converter) that, if needed can translate ANY sample rate into ANY sample rate.

that's OK because at 48kHz your system has a whopping <20uS to get everything aligned again before the next samples show up. install ASIO4ALL and you can see the hooks it can employ if needed. it's a great, stable program, been using it for 15 years I guess.

at higher sample rates your system has to work harder.

[serr]

Originally Posted by drumphil

Windows audio doesn't have a feature to resample devices with no hardware wordclock link to keep them synchronized. Neither does ASIO4ALL.

I'm curious to know how usable the mac audio system is at low latencies when using resampling to maintain synchronization, because resampling necessarily takes time, adding to latency.

I'm also curious to know how or if it accounts for the varying levels of input and output latency between different audio interfaces at the same buffer size. How well aligned do the tracks from different interfaces end up?

Is this feature usable for real time audio, or is it mostly just to allow cheap audio interfaces to be used simultaneously for applications where the above concerns aren't critical, like podcasting with a USB mic, using the mac's headphone out for monitoring?

You can get a very occasional dropout if/when the system re-syncs. I wouldn't do this with live sound. Great post production workaround though.

Different interfaces usually have at least slightly different latencies. You can calibrate for everything based on loopback results and it all stays put. (It isn't random.)

[drumphil]

2 - that means your system takes in data at different rates, that is handled by a device called a Sample Rate Converter (SRC). a higher quality, slower type of SRC is an ASRC (asynchronous sample rate converter) that, if needed can translate ANY sample rate into ANY sample rate.

Yeah, but as far as I know, ASIO4ALL doesn't do that. It does have an option for resampling everything to 48K (because AC97 audio devices on old motherboards usually could only do 48K in hardware), but it doesn't use resampling for drift correction.

You can get a very occasional dropout if/when the system re-syncs.

That's not drift correction then. That's just dropping a buffer when things get too far out of time. If my understanding is correct, the whole point of drift correction is to use resampling to prevent that from ever happening.

If I recall correctly, in older versions of mac os, it was labeled "resampling" rather than drift correction. If you're just going to drop a buffer and resync occasionally, you don't need to bother with resampling. So no latency penalty, but it glitches occasionally.

[Tone Deft]

Originally Posted by drumphil

Yeah, but as far as I know, ASIO4ALL doesn't do that. It does have an option for resampling everything to 48K (because AC97 audio devices on old motherboards usually could only do 48K in hardware), but it doesn't use resampling for drift correction.

as far as YOU know. I tried to explain, you won't get it. if you're satisfied with your level of understanding that's fine with me. I have nothing to learn from you.

[drumphil]

Originally Posted by Tone Deft

as far as YOU know. I tried to explain, you won't get it. if you're satisfied with your level of understanding that's fine with me. I have nothing to learn from you.

Exactly what did you explain?

install ASIO4ALL and you can see the hooks it can employ if needed.

You can see the hooks it can employ?

What exactly does that mean?

Where did you get the idea that ASIO4ALL uses resampling to allow for multiple devices with unconnected clocks?
I want to understand this stuff properly. If I can help someone understand something better then I'll do my best. If someone else can expand or correct my understanding of how things work, then I'll do my best to take advantage of that.

What is your point, why are you making it, and why are you pissed off at me?

[Tone Deft]

it's OK, of all the things that can piss me off the internet is not one of them.

look a few posts up, at the bottom I tried to go into detail about how no two audio sources are truly in sync. there has to be a common meeting point for all the data sources. that common meeting point has to communicate at various speeds because that's the scientific fault in the tolerance of crystals used for electronics clocking schemes.

let's up the Holiday Cheer this year, people need that stuff right now.

[drumphil]

Originally Posted by Tone Deft

look a few posts up, at the bottom I tried to go into detail about how no two audio sources are truly in sync. there has to be a common meeting point for all the data sources. that common meeting point has to communicate at various speeds because that's the scientific fault in the tolerance of crystals used for electronics clocking schemes.

That's digital audio clocking 101. The program I was talking about that combines ASIO drivers still requires a wordclock link between the different audio interfaces.

To be aligned in time, and not just synced to the same clock (which are different things) the different audio devices involved need to be accurately reporting their input and output latency.

let's up the Holiday Cheer this year, people need that stuff right now.

Hey, I'm all for that. I just like to be clear about exactly what we're talking about.

[marlaslangelis]

Just what I was looking for, thanks!

Originally Posted by serr

It's a two step process:

1. Make the aggregate device to combine the multiple interfaces into a single virtual interface.
2. Choose one of the interfaces as sample rate clock master. Slave the other interfaces in the aggregate to the master.

The 2nd step is what loses people I suspect.
Also if you have an interface with no word clock or digital I/O, you're left to try sync over the data connection (USB/firewire/TB/logic board integrated) and this may not give the best stability. Or it may only work at SD sample rates. etc etc

Multiple interfaces all with word clock and properly configured are rock solid stable. Using a digital audio connection to also carry sample rate clock when word clock isn't an option is also usually just as rock solid stable.

[marlaslangelis]

Connecting multiple devices today is called an "aggregate" -- that's what I learned.

I've combined my old Presonus Studio 6|8 with my new Audient iD14 for a current total 6 outputs (expandable to 16 if I had the 2 SPDIFs from the Presonus and the extra 8 from the Audient).

Here's Kenny's video on how to do it for Mac.

https://www.youtube.com/watch?v=NrzcEYzcbXA

[Liquid Fusion]

What about Syncing multiple devices on a PC with Reaper? Not everyone uses a mac.

Reaper > Preferences> Audio> no choice of Aggregate Devoice.

[jerome_oneil]

Originally Posted by Liquid Fusion

What about Syncing multiple devices on a PC with Reaper? Not everyone uses a mac.

Reaper > Preferences> Audio> no choice of Aggregate Devoice.

Reaper only knows about the interfaces the system tells it about. It doesn't build them on its own.

You need to build them using whatever the Windows audio system provides you. That is by the reading of this thread done with AISO4ALL. Use that and build your aggregate, and Reaper will see it.

And do make sure you have the word clock thing synced.

[ssimlai]

Originally Posted by jerome_oneil

Reaper only knows about the interfaces the system tells it about. It doesn't build them on its own.

You need to build them using whatever the Windows audio system provides you. That is by the reading of this thread done with AISO4ALL. Use that and build your aggregate, and Reaper will see it.

And do make sure you have the word clock thing synced.

What about Focusrite and Asio4All?

[bcain]

Originally Posted by ssimlai

What about Focusrite and Asio4All?

I have a Solid State Logic SSL2 and Zoom R8(configured as an audio interface) that I've been using for a while with Reaper running in Windows 10. ASIO4ALL is how I marry the two together and it works.

However, I have experienced some issues that makes me wonder if it's clock sync or buffer related. On occasion a track (mostly recorded through an R8 channel) will have static. Sometimes it's just in spots and other timns the whole track is corrupted. It's happened several times. I have completely disconnected the R8 for now. With that said, I did get some brief clicking/popping in one small area of a track that was recorded through the SSL2. Again, I wonder if it's buffer related? I typically just use the default settings; i.e. Buffer = 512.

[nofish]

This seems to be the new sh*t for using multiple audio interfaces on Windows (not tried myself yet):
https://forum.cockos.com/showthread.php?t=286066

[jerome_oneil]

Originally Posted by ssimlai

What about Focusrite and Asio4All?

I run a MOTU pre-amp through my Focusrite 18i all the time and manage it with the Focusrite software, which is really good. I just use the lightpipe interface between the MOTU and the 18i and the Focusrite. It picks up the 8 new channels coming from the MOTU, and reports it all to the laptop as one interface. That requires no specialized ASIO configuration at all.

I've only done aggregated interfaces on Macs, which makes it easier, but I'd still rather have just the "one" interface to manage in the system's audio setup.