Minimising latency while recording/monitoring audio

Just been reading up on this, I know there are lots of posts already on latency, but a few things I couldn’t totally clarify:

  1. The “adjust for record latency” option is not explained or even mentioned in the manual… As I understand from other threads, this is for when recording midi - if there is any latency, there will be a tendency to “play ahead” - in which case the midi notes will end up placed on the piano roll earlier than intended. So this option compensates for that, effectively placing the midi notes on the piano roll at the moment when you’re hearing the notes from the monitors (eg after latency), not the moment when the midi key/pad is actually struck. Presuming I’ve understood this correctly(?), is this only relevant for midi recording or does it have any application whatsoever when it comes to audio?

  2. Right now I’m recording direct input guitars - it’s nice to monitor thru cubase, as then I can hear the Guitar Rig amp sound etc as I play. However, even with low buffer settings, I find it hard to record dead in time - after all, am I timing the physical impact of pick and guitar string, or the monitored audio, which is however-many ms later due to latency? Obviously reducing buffer size and using “constrain delay compensation” helps, but if I want absolutely no latency I use direct monitoring (ie thru my Fireface 400 mixer) - sucks to have to record a distorted guitar part completely clean, but at least the timing is going to be right! Or is it… are there any other parameters I need to adjust to make sure direct monitoring eliminates recording latency?

  3. There is one other parameter I adjusted, the “record shift” value, which I’ve always left at 0 in the past. To determine how to set this, I activated the click, and then recorded it back into cubase. Zooming in on the audio wave-image, it seemed to be roughly 3 or 4 ms after the beat (or around 200 samples). So, I’ve entered a value of 200 samples in the “record shift” box - did the test again, and the recorded click was now dead on the beat. Presumably the initial latency here was just the inherent soundcard/system latency, and as there’s no way to eliminate this, cubase gives you the option to manually compensate - have I understood this correctly? And with this done - and while using direct monitoring - is there any reason why my guitar part shouldn’t now record with no latency whatsoever? eg when playing it back, I should hear it in exactly the same relationship to the rest of the track, timing-wise, as I heard when I played it in?

I know there’s a lot of questions here, but any advice much appreciated!

With a good audio interface with direct monitoring capabilities (ASIO 2.0) there should not be any adjustments needed, zero latency operations is normal when monitoring the incoming signal through direct monitoring.

However when inserting VST-FX into the recording chain (inserted on the input channel of cubase) all depends on the VST used and how low the latency round trip time for this/that one is.

What also could help is running at higher samplerates, most audio interfaces have lower latency figures with higher samplerates.
I occasionally had midi timing issues which after hard troubleshooting seems to be coming from the midi interface itself, but now-a-days I have no timing issues with midi input.

I’m recording bass myself regular and do not play with inserted FX, I add it later which is doable when concerning bass, guitar recording is another story. Bottomline is that I can’t get the latency low enough and still play comfortable when using inserted VST-FX.

It also depends on how sensitive you are to latency. I have been a bass player 34 years now and played a thousand something gigs. Mostly with big stacks and lousy monitoring, so I’m used to compensate, and anything below 10 ms is for Me quite workable.
Recording midi has not really anything to do with the latency, unless you use a VSTi as sound source or you are monitoring the keyboard without direct monitoring.
MIDI should always be spot on, I have been recording midi drums (Roland TD) for years without any issues.

Strangely I thought the same until I got my Kemper Profiling amp. All of a sudden the rhythm parts were spot on. I had been compensating them too much!!!

Love the Kemper, almost bought one just to take with me when recording guitars. im so sick of line6 pods, after recording with a Kemper for a day.

Actually the Kemper has been absolutely fantastic for bass. The best bass sounds I have ever had.The bass sits in the mix with very little fuss. I have had to do very little eq. There are not many bass profiles out there, but the ones from Ampsound have been really good. In fact it has also been fantastic for jazzy type guitar sounds - my Gibson L4 has never sounded better.

The kemper is great for direct monitoring. However you can do that with Amplitude/Guitar Rig or any other standalone Amp Sim set to a different out than Cubase. I run the standalone together with Cubase, with Amplitude/GuitarRig enabled in the track to be recorded with the same preset as the standalone. My RME Fireface UC enables me to monitor the standalone whilst playing back Cubase. The standalone has negligible latency. If you enable the monitor button in the track to be recorded you can hear the difference. Obviously when recording the monitor is switched off, or if you are overdubbing mute the track.

Thanks for all the replies - I’m not sure how to run guitars thru a standalone Guitar Rig while monitoring (and while presumably having a clean signal go into Cubase) but this does sound like a good solution, so if you say it can be done, I will try and set it up!

Re tolerance to latency, I can generally just about run a buffer of 128 (in the early stages of a project at least, which is when I’m usually tracking guitars), but even with this (and “constrain delay plugin compensation” engaged) I feel I end up with sloppy takes as compared to when direct monitoring - so I guess I am pretty sensitive to latency.

Just to get back to a few specifics - have I done the right thing with the test of recording the click back into Cubase? i.e. is the 3 or 4 ms delay I’m seeing there indeed an unavoidable hardware latency, and have I hence done the right thing in setting a universal “record shift value” to compensate? I know this isn’t really going to make much significant difference, and only applies to where the audio is placed, not what I hear when recording - but just want to make sure I’ve done no harm here!

Also, can anyone confirm that so long as I’m direct monitoring, it shouldn’t make a difference what my buffer size is? eg a buffer of 512 seems best (no pops crackles) so is there any reason to change to 128 when recording with direct monitoring?

My sound card is capable of running two applications at once. Set the input of GR standalone to the input channel you normally use. I use 3&4 as the RME has Hi Inputs here. The outputs should be different from the Cubase i.e. 3&4.

Call up your soundcard mixer and make sure that the Standalone output 3&4 is going to the audio monitor.

F4 brings up the VST Connections window. Create a bus for your guitar - i.e. GR1
Once you have got this working then create an audio track in Cubase the same as your Standalone input - i.e. 3 or 4. (GR1) Insert Guitar Rig. Have the same preset enabled in both versions. Turn on the Monitor in the inspector for that track Playing a note on the guitar should yield a chorused sound of both GRs guitar emulations. It will also give you an idea of the relative latency of Cubase versus standalone. Turn the monitor off and record. You will find that once you have recorded something on this track you might need to mute the track in order to only hear the Standalone.

I hope it works for you. Let us know if you succeed.

3-4 sec delay is quite normal, actually pretty good and you don’t need to set a universal “record shift value” to compensate.At least I never have.

If you are direct monitoring there should be no problem with whatever Buffer size you choose other than how quickly the transport controls work.

Another consideration is to use an interface with built-in fx that would allow the application of those fx while direct monitoring. Some of the Steinberg UR interfaces have guitar fx that can be implemented while direct monitoring, and the UAD Apollo units claim to be able to do this with only 1.5 ms of latency. Perhaps not the exact fx you want to use, but easier on the ears than a dry signal. In both cases, you can record dry while monitoring with fx.

In the case of Guitar Rig there are a wealth of efx you can utilise to get the sound you want.

Many thanks for the setup tips for Guitar Rig, have it working now - routed it out to 3/4 like you said, then sent this to the main monitor output. It still seems to have considerable latency though - at a 512 buffer, it claims to have a latency of 24ms, which isn’t much less than what I’m getting monitoring through cubase?

However, what I have done is activate the Asio Guard - another thing I’ve previously left off. Suddenly, I can run busy projects at a 128 buffer (and may be able to go even lower, haven’t tried yet) - at this buffer size, the standalone GR reports a total latency of 8ms, which is OK I guess.

So looks like this is the solution to tracking guitars without (much) latency, while still having the amp/fx sound to listen to - much better than having to record tricky distortion/fx-dependent parts in while direct monitoring a clean sound! Thanks again for the standalone GR idea.

Just two final questions - are there any downsides to turning on the Asio Guard? I understand fader movements etc might be slightly laggy, but who cares about that - as I understand, it doesn’t affect latency of any audio monitored through cubase (eg in relation to the rest of the track) - can anyone confirm I’ve understood this correctly?

And just to clarify re the “record shift” value - I know 3ms is insignificant, but is there any reason NOT to compensate for this?

Well of course the standalone will be affected if you have a buffer size of 512. However as you say at 128 it is acceptable and the advantage of the standalone is that when Asioguard increases the latency the standalone will stay as it is. I didn’t say this was a perfect solution merely making things a little more acceptable. You could try a higher Asioguard setting and go down to 64 samples.

The compensation shouldn’t be needed if the standalone is working efficiently. Playing in time with Cubase should be enough. Actually when you play live and are not standing right next to the amp you get latency from the speaker to your ears. I use a Kemper Profiling Amp and with this I have little or no latency to be concerned about. However, I only get the effected signal unless I record the raw waveform - and then things get complicated when editing. There is no perfect solution, we just try to create the best possible solution we can.

Yep, agreed - wasn’t complaining about the GR standalone latency, just making sure I hadn’t missed anything! It will definitely be my go-to method now for DI guitars.

I’m still not sure re the Asio Guard though - say I’m recording in with a midi keyboard - depending on the buffer size, there’s a delay between pressing key and hearing note (this being the input/output latencies and any additional latencies from fx processing). When I activate the Asio Guard it tells me it’s creating about 20ms latency. Does this 20ms get added to the delay I just mentioned, or does it definitely not affect this input/output signal?

I’d spend a little time figuring out how it works for you. I don’t know your system, but perhaps it can cope with a lower sample rate if you up the Asioguard setting. It doesn’t hurt to experiment a bit here until you find something that works for you.

I can’t say that I notice a delay with midi keyboards. But then I an not a piano player.

Thanks for clearing that up.

To the OP, some Steinberg interfaces have zero latency direct monitoring combined with effects including guitar amps. I have only used the compression and EQ, so can’t comment on the quality, choice, or desirability of the guitar amp sounds. But it might be enough to make you happy possibly.

(I’ve got the UR28m, but since you don’t really need a new interface per se, maybe one of the smaller/less expensive ones will be all you might need).

:laughing: Haha!

To go a bit off-topic (but with some relevance to latency affecting guitar/bass recording), I notice you’re using the Zoom UAC-2 - is that the usb3 interface, and if so, what’s the latency like and how are you finding it?

Steve.

I’m trying it on USB2 and latency /overall driver performance is just great. My understanding is that figures for USB2 or 3 are almost the same. There are some issues with the latest driver for some people with sample rate switching but I work at 44.1 so it doesn’t affect me.
I really got it because Scan in UK were knocking them out for £125 and I wanted something I could chuck in a bag with a laptop. Sound and features wise it can’t compare to my Audient (which has very similar performance to the Yamaha/Steinberg USB drivers).

However, now that deal is over they’re near £200 which I think is pushing what it’s really worth…driver performance aside I don’t see it’s worth double the price of a UR22. But if low latency is critical for you it’s still way cheaper than an RME.

Thanks for that; wish I’d noticed it at £125. I get on okay with the latency of my UR824, currently about 10ms total in and out at 64 sample buffer size at 44.1 Khz, not particularly low but playable. Do you know what Cubase reports as the in/out latency for the UAC-2?

Steve.

At 64 Cubase is reporting 5.46 in/out combined. (44.1k)