DACs

You can test a DAC with a patch cable#

Loopback tests are trivially easy to conduct: you plug the output of an audio interface back into the input. Then play audio and record what comes back. The difference between the original and the loopback recording is the difference of what the audio interface does. For DACs you need an ADC, and typically the ADC is the harder thing to get right, but if you can't hear a difference, then it would indicate that DACs are a solved problem.

Nobody could hear the switches#

So I conducted a loopback test for YouTube and just asked the question "how many times did I switch between the original and the loopback?" There was no value judgement involved, of which sounds better, just "how many switches". Because if you can't even hear when I'm switching, then all further debate about which sounds better can easily be considered superfluous.

1,367 people took part in the survey and the results indicate that nobody could reliably detect the switches. In the comments under the same video, about half insisted that DACs sound clearly different.

The gear was not especially high end and the signal path used unbalanced cabling with mini jack patch cables, so more or less the opposite of what the audiophile crew would be happy to have in their living rooms. Yet the results indicate that even given that setup, the difference was not audible.

The full method, files and results are in the PDF.

The chip does the heavy lifting#

This was unsurprising given that nearly every audio interface and hi-fi DAC is built around the same handful of chips from the same four manufacturers. Contrary to the Pavlovian response of the audiophile devotees, "output stage tho!", the chip does the heavy lifting and the output stage is essentially trivial. In many cases the chip manufacturer provides a schematic for the output stage in the technical manual.

A flagship chip costs maybe a dozen or a few dozen dollars, not hundreds. Measured differences between competent DACs are in the region of 0.01 to 0.1 dB in the high frequencies, which is typically at least an order of magnitude smaller than what anyone can reliably hear.

Different does not mean better#

Some high end DACs use FPGAs, no oversampling and/or approaches other than what's typical in the mainstream delta-sigma chips. But doing something different does not mean better. A well-designed professional quality DAC is transparent by definition and is invariably priced closer to a new washing machine than a new Ferrari. Paying significantly more for an exotic DAC, when it has already been demonstrated that there is no audible insufficiency in the standard professional level DACs to motivate it, can be explained better by marketing or aesthetics than by technical performance.

If two DACs sound different, one is broken#

If two DACs sound different, at least one of them is not transparent, which means it is broken or badly designed. Such units do exist. Extremely cheap plastic junk exists and poorly designed DACs also exist.

Price#

The sockets, power supply, case, channel count and build quality cost real money, and wanting a beautifully made box is reason enough to buy one, so I do not expect every DAC to have a cost floor on par with a box of cornflakes. But the idea that "you get what you pay for" applies to audio converters, a set which includes units priced similarly to a small Mediterranean house, seems to lead people off the trodden path of common sense and into the esoteric audiophile weeds.

If you feel yourself seduced by expensive shiny metal boxes, stop. Take a breath. Buy a MOTU UltraLite mk5 or something suitable from the pro audio arena and thank me later. The same goes for cables.