Audio Interfaces: The High-End

by Pablo Bellinghausen

We have received a few questions about the quality of different sound cards lately, and so for the last of our audio interface articles we will be talking about the high-end products found in professional recording and mastering studios, and what to expect from them in terms of sound, features and usability.

Sound recording in general has evolved incredibly quickly over the last few decades, allowing for a vast range of products aimed at anyone from budding producers and engineers to professional studios and commercial broadcast facilities. When it comes to sound cards however, the differences in price can be rather baffling: nowadays, a stereo audio interface can cost anywhere from about £30 (Behringer U-Control UCA222) to over £3000 (Antelope Eclipse 384).

Antelope-Eclipse-384-Main

Antelope Eclipse 384: amazing quality and connectivity for a stereo interface – at a price

At this point, and without a good grasp of the theory behind digital audio, the differences can be hard to judge. After all, why pay a hundred times more for a device that ideally isn’t supposed to impart a sound at all, but rather is just supposed to send clean audio feeds into and out of the computer as invisibly and unobtrusively as possible?

As it happens, there are quite a few things that can degrade the sound before, during and after conversion. None of them are intrinsic problems in digital audio and all of them can be overcome, but when lowering costs some corners will inevitably be cut.

Prism-Lyra-2

The Prism Lyra 2: one of the very best stereo sound cards money can buy

The first thing that will suffer when going down the price range is the quality of the analogue stages around the AD/DA conversion itself – from the quality of the mic preamps and the instrument DI transformer to the power and transparency of the headphone preamp. Impedance mismatches at any stage are common and can unpredictably alter the bass and treble responses depending on the device to which they’re plugged in – a problem that explains much of the voodoo in audiophile audio.

Crane Song Hedd 192

Crane Song Hedd 192: Top-notch analogue stages with sound-shaping options

Budget interfaces (and indeed a disheartening percentage of portable and budget media players) will also often lack the required headroom at the output stage to play inter-sample peaks (very common in mastered commercial music, with transients often being way louder than maximum digital values), leading to clipping distortion during playback.

Intersample-Distortion

Digital signal with content above 0 dBFS; possible clipping shown in red

Due to the mathematics involved, effective analogue-to-digital conversion relies on low-pass filters removing all frequencies above half the sampling rate (24 kHz for a 48 kHz sampling rate, for example). At lower rates like the CD-quality 44.1 kHz, these filters need to be extremely steep, which is very difficult to engineer in the analogue world. Arguably, harsh-sounding filters in early converters were one of the reasons why digital audio was at first considered as sonically inferior, and it is ironic that most of the sound quality degradation that some people call “digititis” has actually come from inadequate, badly-designed or underpowered analogue stages instead.

The A/D and D/A conversion itself is performed by a dedicated chip that is almost never made by the audio interface manufacturer itself, but is important not to skimp out on; cheaper models will suffer from inaccuracies in the voltage measurement and an insidious effect called jitter, which happens when all the samples are not taken at the right time. This will cause the sound to become slightly “smeared” or unfocussed, particularly noticeable in hard transients of instruments such as drums or synths.

Jitter

To avoid the aforementioned harshness of analogue filters, converter chips will nowadays always be set at a high rate (a process called oversampling), then using a digital filter to convert down to the desired lower sampling rate. When done right, this operation is completely inaudible, making it possible to get CD-quality audio that sounds as good as the master; these calculations need a powerful chip and good algorithms however, neither of which are always the case.

TI-Burr-Brown-PCM1795-b

Texas Instruments (Burr-Brown) PCM1795: one of the best converter chips available

The last thing to consider is the quality of the supplied drivers, which control the communication between the digital audio stream, and the operating system and recording software; better companies will employ more talented programmers and will devote more time and resources to the coding, who will be able to achieve lower latency and more stability in a wide range of different computers. This can be critical in certain situations where there is no possibility of doing another take (such as in live broadcasts, concert recording or live performance) but even in a more relaxed environment, poorly-programmed or finicky drivers can wreak havoc on a session. Sadly, some interfaces will just not play nice with a specific computer, and short of the manufacturer completely rewriting the drivers the only option left to the customer is often to replace the sound card for another model.

RME-Fireface-UC

RME Fireface UC: RME has a reputation for some of the most stable drivers around

 

Additional Notes

It can be easy to obsess over small differences when wondering what to buy, since the current array of offerings can be hard to navigate. It is however important to note that although the quality of sound and reliability does noticeably improve as one moves up in the price range, even basic models will offer acceptable quality for basic and even commercial work. As with so many things in the audio world, a skilled engineer can do wonders even with budget equipment.

One telling test is to play a great song from the interface’s outputs, send it to the stereo inputs (being careful not to create a feedback loop), recording it, and playing it back. Benchmark Audio is reputed to have done this up to 20 consecutive times without noticeable degradation, but that’s really only to prove a point; If the song still sounds great after one conversion round-trip, then your interface won’t be the thing that holds you back when making and producing music.

Next time we will look at the more practical side of recording vocals for both singing and spoken word projects, with tips that will hopefully be helpful to beginners and more experienced people alike. Until then, happy recording!