If you are looking at buying a new audio interface, this video is worth watching. It’s a comprehensive test and review of a range of home studio audio interfaces that are currently on the market.
Cheers,
Keith
11 Likes
Thanks Keith. I picked up my UMC404HD last year, best I don’t watch this in case it triggers an attack of regrets 
It won’t David.
The UMC gets good reviews and is, in fact, one of his picks, as it was last year when he did this.
I thought that may have changed due to new models and some cheaper prices from other brands, but it’s still up there.
Also, I will add, if you already have an interface, and are happy with it, then I wouldn’t worry. In the first part of the video he goes through his testing methodology and some observations.
Summarising: modern audio interfaces are all very good quality and the differences in audio quality between them is often negligible.
My observation is: modern consumer audio interfaces, pre-amps and DAWs are better quality than anything that even the best recording studios in the world had in the 60s, 70s and early 80s.
Cheers,
Keith
4 Likes
Thanks Keith, that is good to hear, even if my comment was ever so slightly tongue-in-cheek. I’ve had no issues with either of the 2 interfaces I’ve had in my 7 years as a user of such.
2 Likes
So, the takeaway is that I should upgrade to a UA interface.
Got it!
Sounds reasonable. So the internface model is chosen, now to the next, important question: where to get that amount of money?
I didn’t get too far into it to be honest since I’m perfectly happy with my Behringer U-Phoria UMC204HD unit. Also to me, using his own words, he’s a “random person online” as well. It might be better to consider how much you are willing to spend and check reviews from various sites (Amazon reviews can often be helpful, but also from random people) and of course players from this community. So far we have two recommendations for Behringer audio interfaces.
He be unknown, but he does cover most of his testing methods. I will give a great deal of merit to a well defined test method if the method also agrees with what I think it needs to give useful results.
He had a couple I don’t think are all that good, and a couple that seemed ok here. He seems to have a slightly better idea of bit depth and sample rate than most folks, but still not very complete and he doesn’t consider those where they matter - at least he didn’t say he did.
1 Like
It’s a long video, so TL;DR
Winners and losers (video link)
Winner for best features: Lewitt Connect 6
Worst: Moto M2.
His recommendations based on budget (video link)
$200 budget - Behringer UMC204 HD
$300 budget - Lewitt Connect 6
$400 budget - Audient ID24
A reminder that I posted this as a suggestion for anyone who was looking to buy a new interface. If you already have one then, unless you are unhappy for it for some reason, there’s probably no reason to get a new one.
Cheers,
Keith
3 Likes
He really didn’t talk about sample rate and bit depth that much because all of the interfaces being compared support similar sample rate and bit depth (at least 96kHz, 24 bits), so it’s not really a significant factor to consider when comparing interfaces.
Or am I missing something?
Cheers,
Keith
Hi Keith,
Bit depth is part of noise and dynamic range. I would consider this important for a professional but not huge for home hobbyist. Even 16-bit isn’t that bad as long as the analog noise is good, but I’d prefer to go there for the final but still do all the interim processing at 24 or 32 bit depth.
Similarly for sample rate, I’d want a very high rate for anything I was adding non-linear processing to. This will give enough bandwidth to roll off high frequency content produced by the non-linear process so there is no chance for it to fold into the audio spectrum. Again for the final output, we can decimate the digital back to 44k or 48k without concern.
So, 96kHz will give us an octave more bandwidth to roll off the products that would otherwise fold into our hearable spectrum, 192k gives us another octave. A first order filter rolls off at 8dB per octave. Digitally, it is not too hard to generate high order filters, but I am not sure how this was done in something like my Helix. Did they stay true to the analog design they are trying to emulate or did they fudge the digital processing to make it roll off fast and avoid folding? What about DAW software/plug-ins?
If I am talking about something you are not familiar with, let me know and I’ll find a web site that does a good job with what happenes when we do non-linear digital processing - and almost everything is non-linear other than filtering (EQ). I’m guessing you know that though…
I’m very familiar, as I studied Electronics Engineering for my Bachelors Degree, including sample theorem, Nyquist , Fourier transforms, and a short module o acoustics.
I’ve also done a further course on Audio Engineering, and used to be a beta tester for a well known digital audio company.
And I’ve spent most of my life working in Telecoms with digital communication systems where many of these principles have relevance, albeit not always at audio frequencies.
My point is, this video isn’t supposed to be a primer on digital audio. It’s a comparison of audio interface.
And whilst bit depth and sample rate impact analogue audio interface specific ways, given every audio interface in this comparison has the same bit-depth and sample rate, they will all, nominally, have exactly the same behaviour for every interface.
So it seems pretty pointless and irrelevant to compare them.
Cheers,
Keith
I have done a bit of ADC and DAC circuit design in the past.
From that, it is my experience that the DAC filters are usually built into the DAC chip, at least in part (often a few external components are required, but these are part of the reference design).
It’s actually quite difficult to.mess this part of the design up!
The biggest issues tend to be analogue noise from PCB routing, power supplies, or inadequate cooling design causing components to operate too hot.
In the video, he mentioned the convertors and said that he had tested them, including an example, but the results show negligible difference across the units being tested.
Cheers,
Keith
Regarding sample rate, I think it does matter for what happens once you enter the DAW. I didn’t hear anything about it in the video.
One thing I was thinking was a little odd was the re-sampling 100 times. I couldn’t see how this was a quantitative test. This is one that I don’t give much confidence.
I didn’t hear how he did his noise test. Maybe I missed it, but that one seems important to define as interface noise is a fairly important part of getting a clean sampling so it doesn’t show up in the final output. I can think of a couple ways he may have done that, but I would prefer a repeatable measurement style of process.
My experience is with radio design - mostly cellular. The DSP there was often at the edge of its headroom, so we needed to be careful of what we did. I didn’t do those designs, but I did need to deal with the analog result (DAC side) and worked with the DSP folks to clean up problems they didn’t realize they generated. We didn’t always have room for oversampling or heavy filtering on either ADC or DAC sides due to processing speed, latency, or memory depth.
A major part of my work in RF circuit deign, less so once I moved to system design. Cooling 250W of class AB output stage is really important! Even handset output stages need to be cooled carefully so the user doesn’t think the phone is too hot. I even need to work with noise today in a mostly digital job. Noise can be RFI, EMI, and even reflections from discontinuities. Noise is probably a life-long topic. 
I was looking at the other side, the ADC side of the audio interface.
I’ve had a UMC204 HD for the past several years and have been very happy with it!
2 Likes
Because it would be exactly the same for every interface.
I actually thought it was a great test: if a convertor (or the preamps) have some sort of inherent distortion, then every time you do a conversion, the effect of that distortion will multiply.
If, you need to perform the conversion 100 times to vaguely hear the distortion, then it’s reasonable to say that any distortion when the conversion is only done once will be inaudible.
Yes, he didn’t really detail how he measured this. I’ve commented on the video asking him about this. If I get a response, I’ll let you know.
It’s pretty similar. The main difference is you tend to have an analogue pre-amp in front of the ADC, and getting the impedance right here tends to be more important.
He did test the pre-amps, although I’m not sure how he separated the pre-amps from the convertors in the tests.
Cheers,
Keith
Good share Keith, nice to see Behringer getting some positives for a change. My ears are not sensitive enough to hear the quality difference between the UMC Preamps and the Xenyx but hey I am no expert. Buying cheap does not always mean buying twice.
They’re great interfaces in my experience, and scored very well in the video (better than the Foxusrite).
But, as I said, most modern audio interfaces are of a very high quality these days. Any AI on this list is probably indistinguishable from any other on this list as far as audio quality is concerned for most people.
Cheers,
Keith
1 Like
Nor me. I think they have relatively high noise compared to the Midas preamps, but it’s nothing I’ve had an issue with on my XR18.
I have heard it suggested they often put the Midas preamps in because people respect that brand name when they see it on the spec.: it’s to satisfy the gear snobs more than for any real reason.
I would like to see measurements to know how true this is, but I’m happy with my XR18.
Cheers,
Keith
1 Like
This is where I was thinking about how to turn this into a number.
But then, we have the problem of understanding the number! 
What number is a good limit for when we need to care about the change imparted on the signal. So much to be curious about, and so little time.