In a departure from his normal style of videos for Production Expert, Technical Editor James Ivey lives up to the technical title and shares the recent experience of some "scientific" testing on four current audio interface models. 

James was invited to an audio test facility to carry out a number of tests to see just how different a crop of the latest audio interfaces really are.  The units lined up for testing were the Apogee Quartet, Arturia Audiofuse, RME Babyface Pro and the Focusrite 6i6 and range from just under £200 for the Focusrite to around £1400 for the Apogee. 

There were four tests carried out, each of which James describes in the video, and you can see the results of these in the table below for the EIN (Equivalent Input Noise) test, the Round Trip Latency test, Output Dynamic Range test and the Input Frequency Response.  

All test were carried out at 48KHz sample rate. The noise measurements were taken with a linear response. Neither A or C weighting were applied.

The graph below shows the results for the 4 interfaces in the Input Frequency Response test.  The table results show the variation +/- the target curve but the graph shows how the input stage of the interfaces handle frequencies from 20Hz to 20KHz.

Results

Let's start with the results table for the 4 tests. The numbers marked in green show the winner for each of these tests and lighter green indicates second place.  

EIN (Equivalent Input Noise)
The winner of the EIN test was the Arturia AudioFuse with an EIN score of -129dB which is a very good number but when you compare this to the other units which were all around -126dB and -125dB, in the case of the Focusrite, the results for all the tested units were incredibly good and show a very very low amount of internal noise. I would go as far as to say, undetectable in what I term normal working conditions. Where Equivalent Input Noise tends to show itself is when you need a lot of gain for mics like the Shure SM7b or ribbon mics, and as you turn up the gain if the EIN isn't very good you get more and more noise or hiss as you increase the gain.

Round Trip Latency
As we had time, we ran the Round Trip Latency test twice. Once at 48KHz sample rate and then again at 96KHz to see if there was any difference.  The winner of this test at 48KHz was the RME Babyface Pro with a figure of 3.335 ms of latency. Even though the AudioFuse did not fare so well in this test with 5.554 ms this is still an inaudible amount of latency. Even the greatest of players (and as I drummer I can normal feal latency before I hear it) would not be able to "feel" 6ms of latency in a recording system. The 96KHz test did produce a very interesting 2.9 ms result for the Focusrite with all the others within 1 ms of a 3 ms result.  If this test proves one thing it is that there is a good reason to record and mix at higher sample rates. 

Output Dynamic Range
This test I believe needs a little clarification.  The Apogee quartet did very well with a figure of 121.55dB and the AudioFuse at 117dB. The RME Babyface Pro only scores 109dB which we believe could be down to this being the only unit in the test selection that is buss powered. RME do not supply a PSU with the Babyface Pro as standard so we had to use the buss powering option.  Potentially buss power might not be able to offer what the unit needs to operate at it most effective so this could have been a factor in this result. 

The Focusrite result is also interesting. 98.4dB is quite a low figure. It is far from a bad result but we would have expected higher.  It has to be said this is the cheapest unit is the test by quite some margin, but it also should be pointed out that is was the only unit that was not fresh out of the box.  We did download all the very latest drivers for all the test interfaces but the 616 had been around in the facility for some time so we can not guarantee the unit was in grade A condition. Please see the update at the bottom of the article for more information about this test.

Input Frequency Response
Taking a look at the numbers first. The RME Babyface Pro had the smallest +/- variation across the 20 Hz to 20 KHz frequency range of .0225dB. We know RME pride themselves on their "pure and clean" sound and this goes a long way to confirming this.  But even 0.2dB variation is a very very good result.  If we take a look at the swept frequency graph, we see some degree of variation, but the scale of the graph from the 0 line to the bottom of the chart is only 0.5dB. So where you think you are seeing quite a large degree of difference between the test units this variation is again very very small.  Having said that, between 20Hz and 30Hz you might be able to detect 0.4dB of variation but I'm not sure my ears, playback system or room are good enough to detect it. Special credit should go to the RME Babyface for being almost totally flat across the entire 20Hz - 20KHz range with less than 0.02dB variation with no roll off at either the very low or very high end. Very impressive. 

Conclusion

So there you have it. Some actual data that proves something the Production Expert team have been saying for some time. Most, if not all the gear out there these days is of very high quality and will perform perfectly well enough for you to get your creativity on and produce great music or an amazing sounding post project. Now you may say that you like the layout of one unit over the other or you prefer the software control panel of one unit over the other but when we hear claims that XYZ has a better latency figure or ABC has more dynamic range, ask yourself, how much better does it need to be or how much better is it than the competition?

As we have shown in these tests, with these interfaces there is a small amount of variation but the technical performance of even the cheapest unit is very impressive.  Now reading between the lines, could this be because many of the manufacturers are using the same chipsets as each other? How many possible variations of mic pre or pre-amp circuit or A/D converter can there be? I don't have the answer to this but it is something that is worth thinking about. 

If I were shopping for an interface right now, looking at the numbers and where the units placed in our tests, I would be looking at the Arturia Audiofuse or the Apogee Quartet as they would appear to offer a good balance across the entire range of tests. Remembering of course, that the Quartet has 4 mic pres so this is effected in the price by comparison to the Audiofuse.  However, bang for buck the little Focusrite 6i6 still had it's hour in the sun with that stunning result in the 96KHz Round Tip Latency test. It is also the cheapest unit in the test by quite some margin.

So once again it turns out that neither the money you spend on gear or the toys you already have are hindering you from moments of creative genius. The only thing holding you back is you. 

Update - A Response From Focusrite

Following some comments, here and elsewhere, about the testing methods we used and results we achieved, especially on the Output Dynamic Range test, we are giving the brands, whose products we tested the opportunity to comment. This is what Focusrite had to say...

Interface specification measuring is a notoriously difficult topic to cover, but it's incredibly helpful for anyone engaging in audio production to have at least a basic understating of these kinds of specifications. As such, it’s great to see that Pro Tools Expert has jumped in to try and decode the process and resulting specs for its readers.
 
As part of these specific tests, however, the output dynamic range result for the 6i6 did surprise us. Unfortunately, we did not supply the 6i6 directly and are unaware of its history, so are unable to ascertain at this point whether there is a fault present, or whether the discrepancy with our published result is down to the tests not being carried out to the industry-mandated AES17 standard. Using this measurement standard, the 6i6 actually achieves an output dynamic range of 108 dB (A-weighted).
 
We look forward to working closely with Pro Tools Expert to investigate further and also aiding in any way we can to present these kinds of technically informative articles to Pro Tools Expert community in the future.

We have learnt a lot in undertaking these tests and we are looking into how we can do more of these type of tests in the future.