Advice On Audio Interface Purchase

[mavrick102000]

I'm guessing the Gen4 Scarlett at least in the 8i6 model or it's replacement may be coming relatively soon. Based upon the current sale for the next 2 weeks.

I'd still wait a bit after they release the gen4 to make sure there are no driver/software bugs. Mac is usually less likely, but it can happen.

 

[laxu]

Watched Julian Krause's review of the Audient EVO 4 and the latency numbers look better than in that German review of the EVO 8. So the EVO 8 might still be on the table.

Buffer size	Round-trip latency 48 KHz	RTL 96 KHz
32	4.9	4.1
64	5.5	4.5
128	9.8	5.1

That's only about 1.7-3.3 ms more @ 48 KHz than what he got in his RME Babyface Pro FS review. So maybe the German site was using an older driver or class compliant drivers, as Julian's numbers about match what Audient advertises on their specifications.

One plus or minus of the EVO series seems to be that plugging in headphones will mute the outputs in the back. A lot of other audio interface also do this "headphones 1 = out 1/2, HP 2 = out 3/4" type deal but not the muting part. Since I tend to switch between monitors and headphones, I'd probably just use HP 2 for headphones.

 

[mavrick102000]

Watched Julian Krause's review of the Audient EVO 4 and the latency numbers look better than in that German review of the EVO 8. So the EVO 8 might still be on the table.

Buffer size	Round-trip latency 48 KHz	RTL 96 KHz
32	4.9	4.1
64	5.5	4.5
128	9.8	5.1

That's only about 1.7-3.3 ms more @ 48 KHz than what he got in his RME Babyface Pro FS review. So maybe the German site was using an older driver or class compliant drivers, as Julian's numbers about match what Audient advertises on their specifications.

One plus or minus of the EVO series seems to be that plugging in headphones will mute the outputs in the back. A lot of other audio interface also do this "headphones 1 = out 1/2, HP 2 = out 3/4" type deal but not the muting part. Since I tend to switch between monitors and headphones, I'd probably just use HP 2 for headphones.

Different models can have different drivers and different latency numbers.

I think I recall Julian's review of the native 1 & 2 and that was demonstrated in his latency testing. The "higher-end" or model with more features doesn't always fare as well on latency tests as with the NI Kontrol Interfaces.

 

[laxu]

Different models can have different drivers and different latency numbers.

I think I recall Julian's review of the native 1 & 2 and that was demonstrated in his latency testing. The "higher-end" or model with more features doesn't always fare as well on latency tests as with the NI Kontrol Interfaces.

I checked his Audient EVO 16 review and the latency numbers look similar. Which is kinda disappointing for the more expensive interface of course but I guess you buy it for the extra I/O.

 

[mavrick102000]

RTL #'s look decent on them based on his tests. And he's always testing on Windows, which usually tends to be a bit more finicky with ASIO drivers.

Odd that there is no midi though. I guess they're just giving you all of the other ports.

 

[laxu]

RTL #'s look decent on them based on his tests. And he's always testing on Windows, which usually tends to be a bit more finicky with ASIO drivers.

Odd that there is no midi though. I guess they're just giving you all of the other ports.

Yeah one more set of line ins would have been nice too so I could just keep my mics and my synth hooked up somewhat permanently. But at under 200 euros it's perhaps a bit unreasonable expectations.

I'm in no big hurry to buy tho so might wait and see if Focusrite makes a bigger 4th gen unit.

 

[Sascha Franck]

Watched Julian Krause's review of the Audient EVO 4 and the latency numbers look better than in that German review of the EVO 8. So the EVO 8 might still be on the table.

Good to know.
And fwiw, that seems to be a pretty decent channel. Wondering how I missed it until now...

 

[Orvillain]

I'm kind of interested in the Evo 16 + SP8 combination. Might be a relatively inexpensive way to get a lot of I/O. The problem is the drivers. They're gonna be using off the shelf Thesycon drivers, which are usually a load of old wobbly bollocks.

 

[woody]

Yeah I might be interested in the Motu Mk5 too...if it had more mic preamps instead of a shitload of line ins and outs.

There's these odd gaps in audio interface features.

The budget end generally has 1-2 mic pres, which is sensible because that's enough for most. Ok, SPDIF and/or MIDI is also out since few use those, and you get a compact, cheaper unit. Makes sense.

Then you get to say the MOTU M6 which is 4 pres, but no instrument input up front and no SPDIF still. Price is about 2-3x the budget units.

The next step up tends to go straight to "you have a separate mic pre already, right?" like MOTU Mk5, RME Fireface UCX II etc until you get to the rack units which then have a gazillion of everything.

I looked at what Thomann sells and filtered with nothing more than "3-4 mic pres and SPDIF, under 1000 euros". The list is pretty sad:

• Presonus Studio 1810c
• ESI U168 XT (rack size)
• Focusrite Scarlett 18i8 3rd gen
• Focusrite Clarett 4+ Pre
• Arturia AudioFuse Studio

You'd think that the prosumer desktop unit market would have a few more good options than basically dominated by Focusrite. Out of those I'd probably get the Clarett 4+, but I'm hoping that Focusrite would make a bigger 4th gen Scarlett model with the auto gain and whatnot features.

Hmm, the Audient stuff does not register, there?
The ID22, 24, and 44 have optical, (I don't know about the EVO line)
I have the 22, which I really like, two nice pres, insert points, four outputs for either reamping or dual monitors with nice monitor control via their software.
I don't really know about latency, though, because I never monitor through plugins, or the DAW.
I recorded an EP with it, and got really nice sounds.

https://open.spotify.com/intl-de/album/0U5Q5tgpRy81FAhchIIZKS?si=AH5vT5wdR3eGKI-bAbDVkA

 

[laxu]

Hmm, the Audient stuff does not register, there?
The ID22, 24, and 44 have optical, (I don't know about the EVO line)
I have the 22, which I really like, two nice pres, insert points, four outputs for either reamping or dual monitors with nice monitor control via their software.
I don't really know about latency, though, because I never monitor through plugins, or the DAW.
I recorded an EP with it, and got really nice sounds.

Seems Audient has updated these in the past few years. I thought these were like years old boxes so I've pretty much ignored them. I guess they didn't show in Thomann's list due to optical in instead of RCA SPDIF.

The iD44 Mk2 would have pretty decent specs, but it's a pretty big, and frankly ugly box to put on a desk. It's control layout does not seem like it would work well for my desk which is already cramped - with two big 4K displays, a pair of large Genelecs, Hydrasynth Explorer under them, Macbook Pro, Stream Deck Plus etc.

I did discover another large box that looks a bit dated, but is actually also from 2022. The Avid MBOX Studio. The feature set on that one is pretty killer, but after some googling it seems its latency is terrible. How the hell does a company making Protools release an audio interface with awful latency and what seems to be incorrect reporting of latency to the DAW, based on Avid forum posts.

 

[Sascha Franck]

How the hell does a company making Protools release an audio interface with awful latency and what seems to be incorrect reporting of latency to the DAW, based on Avid forum posts.

To drag attention to their larger offerings.

"Want PRO latencies? Get a PRO interface - and pay PRO prices for it, ffs, you damn cheap skater!"

Seriously, the lack of proper audio interface driver development pretty much everywhere is stunning. It's sooo often that companies are using half-generic Thesycon stuff for ASIO/Win and just rely on class compliant Core Audio operation on macOS. Both of which are ok in case the company is using half decent hardware and as long as you only play some virtual instruments and don't rely on audio software monitoring much.
In case you want the latter (and ideally low system load when using small buffer sizes as well - which is quite another reason why dedicated drivers are a good thing), your purchasing options are considerably smaller.

And it's not even looking to get better. Look at Zoom. My UAC-2 is still one of the best of its class in terms of latency - but what about their new, all so shiny UAC-232? No dedicated drivers anymore, just Thesycon/CoreAudio. Same for their pretty neat AMS series. The AMS-22 would be a perfect choice for me on the road, has exactly the kind of connectivity options I need. But again: No dedicated drivers. Sucks big time!

 

[laxu]

Seriously, the lack of proper audio interface driver development pretty much everywhere is stunning. It's sooo often that companies are using half-generic Thesycon stuff for ASIO/Win and just rely on class compliant Core Audio operation on macOS. Both of which are ok in case the company is using half decent hardware and as long as you only play some virtual instruments and don't rely on audio software monitoring much.
In case you want the latter (and ideally low system load when using small buffer sizes as well - which is quite another reason why dedicated drivers are a good thing), your purchasing options are considerably smaller.

And it's not even looking to get better. Look at Zoom. My UAC-2 is still one of the best of its class in terms of latency - but what about their new, all so shiny UAC-232? No dedicated drivers anymore, just Thesycon/CoreAudio. Same for their pretty neat AMS series. The AMS-22 would be a perfect choice for me on the road, has exactly the kind of connectivity options I need. But again: No dedicated drivers. Sucks big time!

I feel like a lot of these boxes are now designed with content streamers in mind. Just looking for YT reviews there's so many of them that are basically evaluating them based on how well they perform for vocal mics. Latency is mostly a non-issue for those uses.

Another annoyance is how difficult it is to find latency figures in the first place as they tend to be buried in forum posts, YT videos etc. Like how hard is it for these companies to at least report these numbers in their specs with a "results may vary because of hardware/software differences" disclaimer?

 

[Sascha Franck]

Ok, talking about audio interfaces. Well, rather talking about my Zoom UAC-2. I finally got everything I need installed to switch my music stuff over from the old Mac Pro to the new Macbook. So I installed the latest Zoom drivers and firmware as they promised silicon compatibility.
Ok, I rather don't type this again, but this is what I just wrote to their support (no fucking regrets!):

"Ok, what is up with this company?
Your socalled "engineers" just FUCKED UP the UAC-2 royally!
I just bought a new Macbook Air and thought I'd install the latest firmware and drivers as they claim to be silicon compatible.
But NO, your folks just ROYALLY MESSED UP!

First off: Round Trip Latency went from a great 4.9ms at 32 samples buffersize to 7.9ms.
What the fuck already - I mean: WHAT. THE. FLYING. FUCK?!?

Then: The fucking constant stream of SysEx data is back! Read: Your device is messing up EACH AND EVERY MIDI data stream with TRUCKLOADS of goddamn fucking SysEx data needed for NOTHING!

And as if that wasn't embarassing enough already, you folks have the guts to still keep those "Logic Troubleshooting" documents online, telling me I should fucking DISCONNECT the device when working in Logic!
What drugs are you on? You recommend disconneting an audio interface in a situation when an audio interface is THE. MOST. CRUCIAL. thing you need?

https://zoomcorp.com/media/documents/Faults_and_workaround_in_Logic_for_TAC-UAC_Series_2_1.pdf

Ok, tell your "engineers" they can shove whatever non-existing diplomas to where the sun never shines!"

In other words: I need to buy a new interface now because these folks are just uber-dumb morons! Done with that company.
Sorry for the vent (well, not really - take it as an advice to boycott these idiots).

 

[Sascha Franck]

Sorry for interrupting again, but maybe it's a warning for some folks...
Because well, it's even worse than I thought with the Zoom morons.

As this interface is pretty much useless for what I want for mobile usage, I tried to revert to the previous firmware. Which seemed to work. Just that it didn't. One should know that the UAC-2 supports two modes: Class Compliant and "true driver support" (it just says "Class Compliant off") that you switch between on the hardware. You need to re-power the unit for that to take place. It always worked pretty fine. Now it doesn't anymore. Regardless of the firmware and drivers installed, it only works in Class Compliant mode. Apparently the most recent firmware update wrote something to a section of the EEPROM that no previous firmware installer ever accessed. Namely the section responsible for the unit to work in two modes.

So, not only that latency is really bad now (even worse under Sonoma compared to Mojave, 7.9 vs. 6.9 ms, with drivers it's been 4.9), the exposed constant stream of SysEx data (which should really only be kept internal) is back, too. Yes, you can adress this via some 3rd party tool (MidiPipe is nice for such tasks on macOS, it's free too), but I rather not have to. And very obviously, I can't even sell this thing to Windows users anymore ("yeah, works fine but please make sure to use an autostarted MIDI filtering utility!" - right...).

Could I possibly expect a fix? Not at all. It's a pretty much officially discontinued product by now.
What's the saying? Buy cheap, buy twice. Has never been more true.

 

[woody]

I feel like a lot of these boxes are now designed with content streamers in mind. Just looking for YT reviews there's so many of them that are basically evaluating them based on how well they perform for vocal mics. Latency is mostly a non-issue for those uses.

Another annoyance is how difficult it is to find latency figures in the first place as they tend to be buried in forum posts, YT videos etc. Like how hard is it for these companies to at least report these numbers in their specs with a "results may vary because of hardware/software differences" disclaimer?

USB audio interfaces don't come with deterministic latency. Latency changes with every restart of the computer.
Thunderbolt has deterministic latency, and some bespoke connections like Metric Halos MHLink might have it.
USB does not.

 

[Sascha Franck]

USB audio interfaces don't come with deterministic latency. Latency changes with every restart of the computer.
Thunderbolt has deterministic latency, and some bespoke connections like Metric Halos MHLink might have it.
USB does not.

WTF? I have performed dozens of latency tests - and that's just utter bogus (unless we're talking about values <0.xms).

 

[woody]

WTF? I have performed dozens of latency tests - and that's just utter bogus (unless we're talking about values <0.xms).

Which might be enough to alter intermodulation distortion between tracks in an audible way.
If you want to recall a dense mix with a lot of hardware inserts (which is what Chuck routinely has to do), this might get a reasonably big deal.

That said, I came to the conclusion that latency is kind of a crap shoot.
So I usually just use the real time monitoring and use my computer just like a tape machine while tracking, forgoing plugins to track through.
That serves my pretty well. It also lets me commit sounds right away. Which is a plus in my book...

 

[Sascha Franck]

Which might be enough to alter intermodulation distortion between tracks in an audible way.

Pardon? How so. Please elaborate.

If you want to recall a dense mix with a lot of hardware inserts (which is what Chuck routinely has to do), this might get a reasonably big deal.

No. Because the "offsets" (if you will) stay stable from boot to boot.

That said, I came to the conclusion that latency is kind of a crap shoot

Pardon? I can replicate my latency measurements to the 0.x ms each and every time. What exactly is a crapshoot there?

 

[woody]

Pardon? How so. Please elaborate.


No. Because the "offsets" (if you will) stay stable from boot to boot.



Pardon? I can replicate my latency measurements to the 0.x ms each and every time. What exactly is a crapshoot there?

Great if it works for you!

 

[Sascha Franck]

Great if it works for you!

Well, it works for an entire industry.

But still, I'd like to learn more about that intermodulation distortion between tracks...