Of converters …

Of converters …

“Digital audio is so bright, brittle and cold as to be unlistenable” … it’s been a long time since I’ve heard those once commonplace words uttered. Fact is those first gen, once cutting-edge AtoD and DtoA converters are outperformed by today’s affordable pro-sumer desktop devices. Its been a quiet revolution, a slow build … but the weakest link of the chain has become amply strong.

Testing and comparing

My ongoing education into converters has been slow and painstaking, with each new product purchased for use in my own room, with each visit to a colleague’s digs, with each loaned/borrowed/finagled unit contributing to the database. Each new experience changing the scope of “what is current, what is expected, what is possible”.

I luckily got to conduct a “Session Trial” (heavily in-depth, comparative-type review) of state-of-the-art converters back in 2010 including Prism, Benchmark Media, Lavry and Mytek. This review was subjective, also with some straight up test measurement, including round trip latency and inverse sum measurement. The sonic results were impressive at the time, but recent developments have moved the bar even higher.

The last year has provided the opportunity for many converters to take up temporary residence at my studio and ease into my workflow. I’ve grown used to the troublesome process of installation, incorporation, adjustment and sorrowful parting, as eventually these units leave me just when I’ve grown used to their effect on my studio’s sound.

I’ve recently used, tested and relied upon new-gen converters from Apogee, iZ (RADAR), SSL, Focusrite RedNet, SPL, Universal Audio and MOTU. I can unequivocally say that there wasn’t a unit in the bunch that didn’t meet my expectations of what I thought was good, even great. In fact, I frequently enlist other engineers and performers to aid in my evals … their insight is invaluable.

Finding results

Fellow engineers all seem to evaluate in their own way, as well as describe their findings differently, but we often find ourselves in near total agreement. Us engineers are used to critical listening and we discern the finest of details … the prevalence of distortion, the presence of noise, the tracking of transients, the accuracy of frequency response. It’s a rare occasion of audio evaluation where specs are not only valid, but accurately indicative. That is, one can read the distortion and noise specs of converters, hear the differences in performance and find a linear relationship between the two. Conversely, frequency response specs rarely show much of anything, not much that the ear can correlate at least.

Musicians, on the other hand, offer a starkly different form of converter evaluation, one that seems more visceral, less informed by perfection and more influenced by psycho-acoustic results. Musicians care about the clarity of less-distortion, but they don’t care about linear frequency response … they’ll usually take punch and euphony over accuracy or sterility.

To my ear …

For starters, 192 k doesn’t seem to be a pivotal parameter. None of my musician testers have preferred 192 over 96, 88, 48 or 44 k … for that matter, my engineer colleagues typically concede that 192 sounds more perfect with some converters (oddly enough, far worse with certain models), but none have actually “preferred” the audio quality, not to mention the massive amounts of disk space. Clinical, stark, sterile … those are the words that keep coming up with 192 k.

(Lack of) noise performance is fantastic these days, with distortion practically eliminated as well. Oddly enough, you’ll occasionally see converters with clean AtoD specs that are significantly dirtier on the DtoA side (one popular unit comes to mind with great AtoD’s but with measurable/audible deficiencies DtoA). Converter chips are implemented with analog circuitry after all.

Even though every pro unit on the market today is presented as “flat from 20 to 20k” they sure don’t sound that way. You’ll find colorful areas in some models frequency response, sometimes only at certain sample rates. Such non-linear color troubles most engineers, while many musicians are enamored with the bumps and dips, loving the emotional results of bubbly bass or sizzly treble.

The state of the art

I’ve confined my work to only 44 and 88k, chosen for either convenience or sound quality respectively. I’ve been on Apogee’s Rosetta converters, but have recently begun trying the newest “standard bearers” (or about as close to a consensus as you’ll see amongst us audio types), the ESS Sabre32 Ultra’s. These chips (as found in the Apogee Symphonys, Lynx Hilo and the new MOTU line) when DC-coupled offer extraordinary THD +N (distortion plus noise) specs, 32-bit filtering, jitter-free clocking and seriously flat frequency response.

I’m finding myself doing less processing in the mix, fewer mix recalls/do-overs and with wider Q’s and fewer revisions in mastering too. The “ever so slight improvement” of these new converters has reinforced my mic choices, substantiated my placements and yielded better results. The differences getting here have been incremental, but we’ve crossed a line and converters may not get much better than this. Let’s all celebrate with our best audio yet!