Found a great way to achieve zero latency for recording guitar/bass in a Native DAW

Hey all, thought I’d share this little ‘discovery’ I recently found that has helped with the age-old issue of dealing with input latency from your audio interface when tracking live guitar / bass through amp sims in a Native DAW environment.

This workaround might be old news, and probably doesn’t apply to people using outboard hardware, such as real miced cabinets, the use of profile boxes like AxeFX, Kemper, or any number of the IR-based pedals that are out there, etc., but I wanted to pass it along specifically in the interest of getting the most out of your software-based amp sims.

I use Nuendo on a Windows 10 PC and typically keep my RME UCX interface’s ASIO setting at a 4096 sample buffer size to maximize resources within the DAW. I don’t track instruments very often, as most of my work in video game sound design and post-production-oriented. However, recently I helped out a composer friend with some live guitars (he is a keyboardist and the sampled guitars he was using wasn’t cutting it), so naturally I ran into the typical issue:

I needed to create a separate project with his music on some audio tracks for playback with another track armed with an instance of Amplitube, ready for me to re-record the guitar parts across multiple lanes to comp together later.

I had to knock my RME’s buffer setting as low as it could go without freaking out the DAW too much (the lowest it could handle was 128, which still threw up CPU Overload Detected warnings repeatedly, even though the signal recorded just fine). I could definitely feel the delay at 128, but it was just a compromise I had to deal with. We’ve all been there, right?

I had originally bought a UAD Arrow interface in the hopes of using it to access the handful of UAD amp sims I had purchased, but that box simply would not work with Nuendo at all. A kind of smack in the face to we Windows users, with UAD really being an Apple-centric company (there’s no excuse for this in this day and age, if you ask me, but it is what it is).

So then NAMM 2023 rolls around and I see the release of the IK Multimedia AXE I/O One for around $130 and got an idea: What if I get this interface and set it up to run Amplitube?

One thing I learned the hard way is that Windows machines won’t allow more than one ASIO driver to be used at a time in a DAW, so I knew I couldn’t use the AXE I/O as an input device for recording while using the RME for playback, as I was originally hoping to do. So - what if I could run the standalone Amplitube app off the desktop and route that somehow?

After I installed the drivers for the AXE I/O, hoping that their interfaces would be input latency friendly, I was shocked to find that it’s ASIO driver had a sample buffer as low as 16 samples at 96kHz, the sample rate I typically run all my projects at (!!!). This was a fantastic discovery; even my RME’s buffer only went as low as 96, which basically makes the DAW unusable.

I loaded up the standalone Amplitube app on my desktop and configured it for the AXE I/O. Just like that, all of the input delay was gone. It felt just like playing an amp at that 16-sample setting.

Yes, that configuration jacked up the CPU usage for the standalone app, but that didn’t matter since it was running outside of the DAW environment; Nuendo didn’t have to deal with running it instantiated as a plug-in within its system. But would it interfere with Nuendo somehow, running separately?

From there, I routed the AXE I/O One’s balanced outputs into 2 of the RME’s inputs with a pair of TRS cables and set up the RME’s TotalMix software to monitor the inputs directly from the software panel. This ensured there would be no noticeable / additional delay introduced from making the round trip through the DAW, and I could hear the amped signal from my monitors instead of having to wear Headphones.

In addition, since the AXE I/O’s interface also has an ‘Amp Out’ output, I recorded that third channel as a clean, mono DI track alongside the amped stereo track - a real flexibility bonus - to be used later with plug-in-instantiated amp sims later on, utilizing the exact same performance, but without having to have dealt with the input latency issues you’d get from recording via the plug-in method.

I pulled up the same project I had recorded to before, but this time with my RME’s sample buffer set to its usual 4096 and re-recorded the parts, making sure the mute the record tracks’ outputs, and simply monitored the stereo input signal.

No latency or DAW performance issues at all this time; this streamlined recording process was a real treat! Initially I wondered if Word Clock was going to be an issue since these two devices were running separately at 96kHz, but fortunately there was no issue there (the AXE I/O doesn’t even have a Word Clock connection anyway).

Nuendo’s built-in latency compensation recorded the parts perfectly in sync with the music and that was that! No CPU Overload issues due to a low buffer setting on the RME, or ASIOGuard needing to be switched off to help lower latency, etc. It just worked.

In the end, I believe my guitar’s round trip was less than 5ms from strings to ears, and I couldn’t detect any audible delay at all. Rather comparable what I had to get from UAD interface, quite frankly.

So - I suppose the point of all of this is to say that you don’t really need to get into the UAD audio interface world for zero-latency instrument tracking using it’s built-in DSP, as I was initially thinking I needed to do. And now, with this setup, it opens up a lot more flexibility with having access to many, many non-UAD playable amp sims that are out there on the market.

Anyway - it’s a long post, but I wanted to share in the hopes that it helps someone else out as it did for me. At the end of the day, for only $130, it was a pretty inexpensive investment to get zero-latency guitar recording without DSP in a native DAW.


This is really interesting, Rodney! I’ve been thinking of ditching my Apollo and this led me down a similar rabbit-hole. The idea of running a standalone solution at a lower buffer and routing through is something I haven’t tried in years, to be honest. But I’m eager to revisit that idea.

1 Like

Yes, I have found it to be a great solution. I’ve really been enjoying the standalone Neural DSP amplifier / cabinet sims lately. More recently, I’ve built a guitar pedalboard that terminates in a Strymon Iridium amp sim pedal, with a Radial Engineering Shotgun guitar splitter in the front end, so that I can simultaneously split the guitar’s incoming signal between the pedals on the board and the AXE I/O One to capture the best of both worlds.