Actually FURIOUS.. Need help, Steinberg PLEASE answer, AXR4 tbolt

Yes that’s true, you can still use the on board DSP FX though of course, must not have explained myself well :smiley:


Hi Norbury, so you are saying that companies added their own standalone console app mixers for mac os, which doesn’t have asio?

Well fair enough. Mac OS uses core audio… and it makes you wonder, why doesn’t UA offer a DM asio mode on windows with their DSP effects? A bit of a missed opportunity if you ask me.

I hate to play devil’s advocate, but core audio is one reason I choose mac over windows… It absolutely crushes asio… Obviously there’s a way to do it as Apogee and logic have dsp integration with core audio… regardless though, if we take asio DM out of the equation, yes, that’s a good feature… however with core audio, I can have multiple apps open using the very same outputs, at different buffer sizes, and even sample rates!
Right now I have PT open at 96K, 128 buffer and seamlessly switch to logic at 44K, 32 buffer just using the inbuilt audio on the macbook. That’s core audio. Secondly, you can make aggregate devices, which is basically combining multiple devices to be seen as one. So… no comparison for me…

Steinberg are the only developer who won’t do native core audio and still have an asio wrapper for Cubendo…

It’s really made me think… Cubase 10.5’s performance has finally outpaced Logic in around 80% of real world use cases… imagine if Steinberg used native core audio instead of having a wrapping layer to asio… But I suppose maybe that’s too difficult to keep windows and OSX versions parallel.

Unless I am mistaken, I think last time I checked wavelab that used the asio wrapper as well, but I have no idea about V10.

Anyway, for arming tracks and monitoring them at 32 buffer with effects, Cubase 10.5 is now the best performing DAW on OSX, with Logic 2nd, PT/Reaper third. It’s come a long way.

But at 128 buffer, PT is king still. So this is the way I see it… I use DSP for monitoring and 128 to play VI’s, best of all worlds kind of thing. Even if I switched to Nuendo full time, I’d still want the same sort of setup… keep the main buffer at 128 to get an overall balance of good performance and still low enough latency to play VI’s. This is why I asked about the output latency of the AXR (the output is the one that determines realtime VI playing latency with a midi keyboard).
If it’s no more than 4MS for the thunderbolt one at 128 buffer, that will be fine. Anything up to 5MS really, max, is OK for VI’s. Maybe not for drums though but you get used to it.

External input monitoring, well, that’s another story, that’s for sure! 128 way too high for 99% of vocalists, and the external synths I use already HAVE d/a latency of their own as well as DSP latency… so I like keeping it as low as possible. Believe it or not, the apollo’s DSP mixer is also way too high when you use the MK2 UA effects, which they don’t mention anywhere pre buying. You have to use the MK1 effects which have been outdone by native stuff these days.
Luna, the revelation UA keep sprouting about, will be subject to the same input latency as all apollos, and VI’s are at a fixed buffer of 128 that can never be changed. I know that’s way OT but i thought someone reading might find it interesting, and not get bedazzled by UA’s marketing spin like I did.
This is also why I am trying to find out everything FIRST this time instead of buying something without all the facts. If I go out and buy an AXR4T right nw and it can’t do what I need, then it’s entirely my fault… I have to be patient and get the info I need.

Ed from Steinberg support (really nice fellow) asked me for my questions via PM but he hasn’t replied yet. I keep hoping he will get back to me with the answers!


:smiley: core audio is actually less efficient needing 3 kernel calls to ASIO’s 2, so ASIO technically will always outshine Core audio in performance. I agree ASIO doesn’t do aggregate devices but in 35 years I’ve never needed to use that feature so I’ll take better performance :smiley: Windows ASIO devices can switch between apps no problem if you have multi client drivers, all my steinberg units have had this so again it’s not a limitation I’ve come across. ASIO4 all will give low latency drivers for on board sound chips on a laptop for example if you want to use the on board sound. I’ve done gigs using this playing a VSTI rack on my HP elitebook and on board sound from the output jack :smiley:

Latency and monitoring is indeed a thorny old problem :smiley: and…sometimes the numbers don’t match the experience either.

Cubase is one DAW that has a single button ‘constrain plugin delay compensation’ which is very useful not sure how other DAW’s get round this.

Even if you have 32 buffer setting, if you’re monitoring through software and have say a look ahead limiter or your master bus…that will give you huge latency with monitoring. Hit the constrain button and bang you’ve bypassed any latent plugins, this is another thing people forget when working :smiley:

anyway, hope you get your answers and end up with a system you’re happy with.

May I ask why Nuendo and not cubase? I used to have both up to V6 then decided Nuendo became too post orientated so just carried on with Cubase. I’ve had no problem doing my movie scores in Cubase.


cause i want to get into some video and post work.

I don’t mean to be rude, but i am not a noob re latency… I have been campaigning for other DAWs to offer a version of CDC from Cubase for, well, a decade, to no avail… I know exactly how it works.
I am so sensitive to latency, that I discovered all the problems with UA apollos, and uncovered all their lies (that they still sell their interfaces on today), entirely by ear… and everything I questioned their support about is 100% confirmed with no workaround other than using the older 0 additional latency plugins (the ones from 2005) to monitor with. I have no problem calling them liars, as right now they have an ad for their fender plugin saying you can track through it with NO latency. Even after everything I have discussed with them they still continue the BS…They use the word “NO”. Which to me means 0. The fender plugin has 131 samples latency on it’s own (from memory, or 121, something like that), on top of the base 2.3 RTL at 44K of apollo. So that’s almost 5MS RTL, without ANY compressors, EQ’s or possible desired tape plugins. Native is much faster.
This is why I am trying to find out if AXR is actually like 2ms through the effects or much higher like apollo is.

Logic goes one better btw re low latency recording - they call it low latency mode and you click a button, same thing as Cubase, but you set the threshold. For example I can allow all plugins with latency, but below 1MS “total” chain latency, to function. This means I can still monitor with softube plugins which have 4 samples latency, which Cubase disables with CDC. I only use a zero latency safety limiter on the master bus (blue cat protector 2) and before that, the glue in 0 latency mode to mix into (all oversampling disabled so it’s true zero latency).

If I was using Logic still though, or even Cubase, my options would be broader as I could keep latent plugins on the master and use their low latency modes at record time… I do miss this a LOT in pro tools. PT has a low latency mode but it disables ALL plugins… not the same thing.

I have never, ever seen a multi client asio driver use the same outputs at different buffer and sample rates… I have seen them use different outputs on the same interface… so… who knows… regardless, I use aggregate so that’s moot.

Well, how am I supposed to get the answers if no one with an interface is willing to help? A simple loopback test using the FREE Ceentrance utility for example, without cubase loaded, just the AXR through it’s own DSP mixer app going through the vintage EQ and comp?

Or what about someone trying just one track with native plugins instead of DSP?
And if that can’t be done, can it be switched seamlessly to native mode, just so someone may use a native guitar amp, then once the guitar part recorded, switched back to DSP for all the other inputs?

No point wishing answers come my way if no one is willing to spend a few minutes trying things out. I guess it’s a NO from me for now, as even steinberg is not answering me… so… i’ll be on my merry way and maybe even buy one HDX card to get started…
Ultimately I have to be honest… I doubt I could live without the way PT does clip gain anyway… so maybe I am deluding myself. I do love Cubendo’s midi, chord track, drum editor and midi inserts… it’s all midi stuff I miss when in PT… but the way it handles audio is great…

The other thing was the matter of cost… PT ultimate with all the post features is very expensive to maintain… like 400 USD a YEAR… and around 2 grand to buy… vs 400 for Cubase crossgrade to nuendo… so there’s that. Currently I am on PT Vanilla… the upgrade price is like 1800 USD. LOL. And the DSP cards are about 2.5K each real world pricing. Then there’s I/O!
But if Steinberg themselves aren’t willing to give me the answers to what I need, I am at a loss as to what more I can do from my end… maybe they just don’t know, even about their own interface!?!

I didn’t mean core audio was more efficient than asio head to head… Steinberg are using a wrapper… How can THAT be more efficient in this case? That was my point…They are converting core audio TO asio… isn’t that adding extra layers? I know asio performs great on windows… and I myself tell everyone about asio 4 all which I use on my bootcamp partitions :slight_smile:


I didn’t get to the studio today and I’ve been installing a new machine over the last week so not had time for anything.

I’ll do the test tomorrow for you if it’s quick and easy :smiley:

do I need to physically connect outputs to inputs?


oh… thanks so much!

Yeah, just connect output 1 to input 1… that’s all you need to do… Put effect on input 1, run the utility… bob’s your uncle! That will answer one of the major questions for me.

I would then ask you, how easy is it to disable DSP mode to temporarily have the device in native mode, to as I said, use a native monitoring plugin I might need, then once that track is recorded to audio, flip it back to all inputs being through the DSP mixer?

If there’s no hybrid monitoring, this could be a workaround.


TNM, I can’t find that RTL program anywhere. The web site doesn’t have it on there anymore and I can’t find any external downloads or any other tools to do it after an hour of searching.

Any ideas?


Spoke to Vin @daw bench and he recommended a utility.

The RTL of AD/DA going from an output to an input (loopback) @32 samples is 4.6ms. adding the DSP FX makes NO DIFFERENCE to this delay at all.
If you look at the screen grab you can see there’s an EQ and Comp on the tested channel.

To bypass DSP zero latency monitoring you just un-check ‘Direct monitoring’ in the studio/Audio device tab.


That’s absolutely ridiculous… Thank you for doing this… sorry I was asleep, just woke up.

I’m 100% out. No point answering anything else.

I have singers even who won’t be able to work with that, they are too sensitive to comb filtering.

Suddenly apollo seems extremely fast. And it definitely wasn’t going through the DAW, but the device’s DSP mixer only? wow.

Apollo I can keep at 2.2ms total RTL and use at least all the MK1 plugins and that doesn’t change at all.


PS… wow… i just realised apollo with two of the mark 2 latest plugins is still faster than that.

I can put an oxide and 1176 on every single channel and have dsp to spare… with one apollo. Yet this steinberg interface is the same price… so it has neve modelling (remember the neve “mode” on the inputs is 100% software modelling), UA has unison. I have some absolutely incredible takes using the neve unison pre on the UA… I am not going to give UA a hard time about latency again after this. Cheers.

TNM, I work with the best musicians in the world, (literally) and there’s no latency when monitoring trough the hardware DSP has you can see, the hardware adds none just like RME. This is RTL plus the driver. The driver doesn’t come into play when monitoring. I think you’re misunderstanding how it works. If there was any noticeable latency the session musicians I use would complain immediately. I had the Head of Warner Bros records sat next to me during the recording of a number 1 album and that was using a UR824 which has twice the latency and again because of the way they work, there’s no latency in monitoring, it was a huge 96k project and I had to work at 1024 buffers so the latency was large, didn’t matter thought :smiley:

For me the sound quality and total integration is what this interface is about and it’s TOP QUALITY in sound build and usability.

the USB AXR is just a little below the RME babyface pro in the table of RTL on the DAW bench Charts. Again no one ever complains about REM total mix latency :smiley:

If I wasn’t using steinberg software then the unit would be not as attractive as it’s the integration and ease of use that makes the difference when your on the front line with the Head of WB sat next to you :smiley: LOL

good luck with your music and health, health issues really put things into perspective :smiley:



1 Like

hi well then I think we are both misunderstanding each other LOL cause I thought you were giving me the total latency through the DSP, without the daw loaded. I actually said that… I asked to just measure it with the console app open, and no DAW therefore no asio driver activated, to know the monitoring latency. I really did, but I must have been confusing the way I asked it, i apologise. 4.6ms through the DAW is perfectly in line with other good USB stuff and just fine. I can find you the utility if you can’t find it, you are on windows correct?

EDIT… misunderstand how it works? Condescending much? It really doesn’t matter who you work with, I have one of the deepest understandings of dsp monitoring systems around, having used them since creamware pulsar/pentium 1 days… I know precisely how it works.
And btw there is no such thing as zero latency… imperceptible, maybe. But all I am trying to find out is what that figure is (through the dsp mixer) so I can decide for myself. I know what outboard I use and what I can feel when playing that makes it uncomfortable to me.

Yes, I did ask for the “through DAW RTL figure” in my very first post, in a separate question of the 5 questions. This was to know how high the latency is if I needed to switch to native monitoring, since the DSP fx on the device are so limited. This is an entirely different thing and you answered that many posts ago already (thank you), so I presumed your latest reply was regarding the straight, non asio DSP figure which is what I asked, which is why I said “that’s way too high”. It actually seems you misunderstood me, but I have been continually polite because you are the only one willing to offer any help… That said, you have been condescending twice now, so don’t bother. If you can’t see how you are acting superior and looking down on me, then, I can’t help you see that, you need to do it for yourself. If I am willing to help someone with something (and I always am if I have access to what they need), I read what they ask and don’t belittle them.
Steinberg obviously don’t want my money since they aren’t answering either and suddenly i have deafening silence on my PMs, so that’s pretty much made up my mind for me. I don’t need to chase a company to give them thousands of dollars of my money, it is their duty to provide the necessary info. They were all reasonable questions. Why are they so scared of releasing the figures if it’s “no latency”? LOL. I already KNOW that’s not true, so why not just tell us exactly WHAT The latency through the hardware monitoring IS so the customer can decide?
Trying to get detailed performance info on this interface is like trying to squeeze blood out of a turnip. It’s very disheartening.

TNM, I’m sorry if you think I’m looking down or being condescending I only mentioned the fact of who I work with as a benchmark and context to say in real use in my working environment then there is to all intents and purposes ZERO latency when using these units in the way they are intended.

Again , I’m sorry I had no intention of being condescending. I went out of my way to try and help. I’m pretty busy here so even finding the damn latency test tool took me all morning and a call to TAFKAT to find the latest one.

I’m sorry no one else has helped. what more can I say?



Hey there,

Seems this has gotten a bit heated. I felt I’d chime in. I own an AXR4U (usb c version) as well as an Antelope Goliath HD, and I have a commercial studio which runs PT HDX1 with Merging Converters (the best available imo based on rigorous shootouts).

The latency on the AXR with DSP is very close to zero. However, there are two “DSP” bars in the control panel. One appears to be for the 276 comp and neve style EQ. You can use two total instances before it’s full, so you could use an EQ and Comp on one channel, or an EQ on two different channels, etc.

The other DSP handles the morphing channel strip, and seems to be able to do about 10 of them (so you could have a more vanilla EQ and comp on quite a few channels. It may also handle the reverb, which if I’m remembering correctly uses about 1/4 of that chip and as an aux effect, remains the same no matter how many channels you send through it.

Steinberg recommended the USB-C version to me and said they had had issues with Thunderbolt drivers on Win10, although that may be worked out at this point. Sadly, I can’t chain two of them as I probably would if I could, but I am using a Win10 PC and don’t want to struggle with Thunderbolt on it at this time.

I find the sound to be excellent and better than most out there. The latency through the drivers / DAW is higher than I’ve experienced on RME interfaces, although much lower than very inexpensive interfaces I’ve tried long ago like Presonus and Focusrite.

I also have a good amount of experience with UAD, including the Apollo X16 (their flagship). I’m not that impressed with the sound of conversion, and my Goliath HD beat it easily in a dry recording shootout on many instruments. I haven’t shot out the Goliath and AXR4U.

All of these companies have fanboys and haters. As far as customer service, Antelope gets a bad rap, but has been FAR better than Avid to deal with, and somewhat better than Steinberg (although I haven’t had to deal with Steinberg as much). Avid has been a nightmare and currently my PT Dock just doesn’t work at all, and I had a console designed around using that, so my whole life is at a standstill because of Eucon and Avid being absolutely horrible in so many ways.

So if you want a lot of channels, I think a Goliath HD could serve you well. They have roundtrip latency of 0.6ms on Thunderbolt, OSX and Win10, with an astounding amount of DSP. You can run 16 compressors and EQs of EACH type, and there are tons of them. You can literally run over a hundred plugins with no latency.

These plugins don’t really come into the picture when it comes to mixing, so you have to just consider them available for headphone feeds only imo or it will be frustruating. But you can create 4 independent stereo headphone mixes with reverb and essentially unlimited effects with no latency. The plugins are comparable to UAD in quality.

Antelope is kind of a bad company as well and updates products way too often, devaluing the used price. They will update one tiny little thing like the monitor output converters getting a 3% increase, and then release that as a new product, making it seem like their stuff is throwaway to them. However they do offer a 5 year warranty on hardware which is excellent.

I’m also curious if you were to chain 2-3 AXR4Ts, would you have 3 reverbs, or at least be able to do 6 of the analog emulation plugs and additional channel strips.

Currently I’m between keeping the AXR4U and Goliath HD for home. I got the AXR for the Nuendo integration and 352.8k sample rate support. I wanted to experiment. I’ve determined it’s really tough on the computer and so many things become unavailable (collaboration, 80%+ of plugins, transmitting audio over AES/ADAT, etc) that I’m going to run 192k, which makes the converter decision tougher.

Both of these interfaces sound very very good and imo significantly better than anything UAD has ever made. The preamps on the AXR are definitely better than the Antelope, but the converters I’m not sure yet. Antelope has higher specs and lower noise floor, but Steinberg offers 32 bit integer (only interface that has this as far as I know and I haven’t tested enough to see if it makes any difference whatsoever).

The other big deciding factor for me is if I want to have like 7 times as much I/O and leave everything plugged in, the Antelope is a better option for me, while I’ll need to incorporate a patchbay with the steinberg.

The DSP and routing on the Antelope stuff has a bit of a learning curve, but it’s really pretty easy if you watch a few videos and it’s very flexible and powerful.

Let me know if you have any questions.

Oh also, AXR latency through effects (which are quite limited) is lower than Apollo I’m pretty sure. Antelope is much lower than Apollo (best on market including HDX).

:smiley: nice ‘review’ and i agree with you on your points.

After using mine now on a few projects I’m completely knocked out on the sound of this thing. the AD is just the best I’ve heard in this class, it’s also built like a tank and uses really high end components.

Yes, the latency isn’t the best but with the zero latency dsp monitoring it’s a moot point anyway. i can run projects @32 sample setting and the other day did an entire project from tracking to mixing using a lot of Acustica Aqua plugins @64 samples the whole way through.

there are a few integration quirks need sorting out but Steinberg/Yamaha are aware of them and will be dealt with at some point.

all in all for me it’s a fantastic interface, and on sound quality and build quality I don’t think there’s anything to touch it in the price bracket.