New Discoveries in Digital Audio (Harley / Guadalajara, TAS)

A blog interview in the-absolute-sound (2026-09-01): editor robert-harley questions javier-guadalajara, co-founder and CEO of the Spanish high-end maker wadax, about distortion mechanisms the company says it has found in digital playback and the technologies it sells to fix them.

It lands directly on this spoke's analog-vs-digital through-line, and on the side the spoke's two AES primaries do not occupy. Where meyer-moran-adda-loop and digital-filter-audibility argue about conversion transparency with double-blind tests, this source argues that the important differences are real, audible, and not measurable at all.

The claims

• The originating experiment. Developing the Atlantis Transport, Wadax built a real-time SACD/CD ripping tool. Swapping only a one-foot USB cable — the ripped data bit-identical — changed what Guadalajara calls the "sonic fingerprint." He is explicit that this was not framed as better or worse, only as different.
• Intra- vs inter-process. Wadax splits its research into effects inside the DAC chassis and effects arising between devices: the analog path outside the DAC, the power-supply and grounding ecosystem, and the incoming digital interface meeting the DAC input stage. The larger remaining gains are claimed to be in the second group.
• Digital Waveform Control (DWC). The shape of the digital waveform, not the data it carries, is said to influence conversion. Their Reference Server exposes rise time, droop and receiver sensitivity as knobs the listener adjusts by ear.
• MusIC, a feed-forward IC. The analog output stage's transfer function, nonlinearities included, is modelled; the inverse is applied in the digital domain before conversion, so the nonlinearity cancels on the way out. The stated analogy is feed-forward linearisation of RF power amplifiers in cellular base stations. Guadalajara says the precision required makes it complex and expensive.
• Architecture matters less than the system. R-2R versus delta-sigma, filter choice, output- stage redesign — all said to matter less than inter-device interaction. A sub-$1,000 DAC can measure very well and still, in his account, not sound like a high-end system.

The sentence the page turns on

"these effects are largely invisible to traditional measurement methods and instrumentation."

That is an unfalsifiable move, and naming it as one is the point. It is the position analog-vs-digital has always had to argue with, stated more plainly than any source here has stated it: the difference is real, you can hear it, instruments cannot see it, and the remedy is for sale.

Compare what the spoke already holds. Meyer and Moran ran 554 trials and reported a null. Stuart and Craven isolated a stage and reported discrimination. Both can be checked, argued with, and re-run. This account cannot be — not because its author is dishonest, but because he has placed the claim outside the reach of the only instrument that could settle it.

Tier and weakness

T3 — vendor claims about the vendor's own products, in trade-press packaging. No listening panel, no blinding, no instrumentation trace, no third party repeating the cable experiment. The USB result has the exact shape that sighted, uncontrolled listening produces reliably: one listener, one variable he chose and knew he had changed, with a product line resting on the answer. robert-harley asks nothing adversarial anywhere in the piece.

The feed-forward IC is the part that survives skepticism best. Pre-distorting a signal by the inverse of a modelled analog transfer function is ordinary engineering with a working precedent in RF, and unlike DWC it is falsifiable at the output with a distortion measurement. Wadax publishes none here.

Related

analog-vs-digital · meyer-moran-adda-loop · digital-filter-audibility · sample-rate · wadax · synthesis