More is more. That confession pains when I'd really favor more simplicity for fewer boxes. But reality chomped down hard. Yum for reality. Ouch for me. What am I on about? On the remaining difference which my Chinese SW-8 left in place between local and cloud files that were equally pampered by Audirvana Studio. The Taiwanese S10 deleted most of this final difference but not all. Cancelling the last bit for what I finally heard as parity required the serial offense of two switches, first my own, then the loaner. Let's unpack this in baby steps by using the title track of Jasdeep Singh Degun's poetic Anomaly album.

The direct connection of router to iMac suffocated all the misty air and subtly textured energy which surround the opening string tremolo that itself underlies the channel-panned sounds. That faint but key milieu came into far greater focus with the S10. Where the local file still won was in the amount of air which surrounds that shimmering string pedal with audible space. When my own switch preceded the COS, the full dose of oxygen restored. Throughout the track the sympathetic strings of Jasdeep's sitar go off like minor buzzers. Router direct rendered those decidedly glassy as though the microphone was much closer to them than the rest of the instrument. When the S10 entered, the sympathetic strings obviously belonged to the same instrument. Their timbre restored. It's only in the final sense of recorded micro reflections which balloon fractional halos around sounds—that's back to audiophilia's concept of airiness or discrete image bubbles—where the S10 still appreciated a little help from my SW-8 to pull even with the local file I'd bought many months earlier through the same Qobuz Sublime account I now used for streaming.

Just for relativity's sake, let's pin poxy numbers on my little exercise. Call the full difference between router⇒Core Audio vs local files⇒Audirvana 10. Routing modem feed through Studio diminished it to 4 i.e. more than half. Atop that, my own switch lowered it to 2, the COS to 1. Finally LHY + COS = 0, meaning equality with the playback of local files. Obviously my starting point of 10 was itself only a fraction of the overall sound. Clearly we're chiselling away at small differences. Think peeler-shaved carrots in a dressy salad. While those curly peels got skinnier over my Qualio IQ speakers, they remained chunkier over HifiMan's Susvara. That's back at applying sharper res to a shootout between subtleties; of not bringing a pocket knife to a gun fight. While on that, a small cut of criticism goes to both switches' absence of any frontal power button. Both insist that we fumble 'round back. True, the vast majority of users won't care. They're 24/7 connected. It's only contentious outliers like us who enjoy to physically disconnect from Big Brother's constant surveillance when we needn't be 'on'. A switched switch makes that more convenient than unplugging an Ethernet cable from behind a computer; and a switch with a frontal power switch still more so. That was the one and only paper cut I could nick the COS with.

For another music sample, follow Karim Baggili's evocatively simple "Zen Zen" to the 2-minute mark. Suddenly more space opens behind the solo oud. That deepens the counter ambiance which by 3'30" hosts what might be a synthesized theremin. Getting the COS involved improved the specificity of those background layers. Whilst the meat of the dish remains the oud, some seasoning and sauce go to the ambiance.

That shimmer of spaciousness from recorded reverb was where the S10 roamed to improve over just Audirvana Studio though—this must repeat to stay honest—the software player was the bigger difference maker. Even if you're a hardware junkie who gets more personal chuff from a physical object than invisible code, I highly recommend that you don't forget about the latter as you investigate your options of graphic user interfaces. Those not only access your cloud music but in the background run their own audio OS or signal routing with more or less tuning and sound-optimizer features. HQPlayer gets super nerdy for example.

So my private takeaway from this assignment was twofold like a classic pincer attack. Circumventing Apple's 2023 Core Audio as I'd bypassed iTunes before remains key. A quality network switch cum noise blocker cum reclocker like the COS Engineering S10 can then close the remaining gap to local files. My inner artist who wants to see musicians get paid more than streaming's criminal pittance twists into knots saying so. Yet that greyed chap can no longer claim that buying digital albums is a must to enjoy the best sound. One can stream them off the cloud and arrive in the same place. It simply takes more than intoning "enterprise-level bit-perfect no-packet-dropped" Ethernet protocol. There's more to it than getting all packets dispatched to our DAC. Does it make sense to still strap an external master clock to a network switch? Why not build the best clock for the budget into the same box as COS have done? That shall remain a question for another day until a Silent Angel GX/NX combo drops in Eire. It'll more than triple today's entrance fee. For that all greedy mitts (should!) expect substantial benefits; like four weeks of paid vacation from all audiophile worries. For today I'm simply happy to report that the COS S10 + Audirvana Studio combo is my current discovery for achieving parity with playback of locally hosted files.

PS: If you were quietly grumbling that my two music samples were so structurally basic, that was their entire point. Rather than feel overwhelmed by a school of whales, we can focus down on the plankton. Once its contributions are sorted, we can progress to more complex loaded fare. Now we know where to look as it were. At least that's how it works best for me. Start simple. Get the lay of the land. Then advance to more complicated challenging stuff until all hell breaks loose.

PPS: To think about this product category properly, I propose renaming it. Instead of network switch, think LAN distributor. Once we do, we see an obvious parallel to power distributor. With it come all the same potential benefits of UHF noise filtering, even isolation between plugged-in components so they don't share each other's self-generated noise out the back door as it were. Power distributors span the gamut from Electric Avenue plastic bars to elite Ansuz/Furutech types with built-in passive noise filtering. Likewise for LAN distributors. They could be as basic as a Computer Emporium giveaway; or as upscale as today's. The core functionality is identical. Packets go in, packets go out. The area of differentiation is what accompanies this traffic by way of noise which worms its way into subsequent signal routing then processing. As seems to be the overriding theme of high-performance digital these days, wherever the conversion process encounters less extraneous noise, results improve. As to why that's so, engineers from Antipodes to Innuos have plenty of hard tech answers. It's their area of expertise so I refer you to them and their colleagues elsewhere when the nitty gritty of it rather goes beyond my modest paygrade. But consumers needn't be tech nerds to benefit. Simple comparison listening will tell us all we need to know.

PPS II: Once the above published, I asked for the usual call-tag pickup. "Do you mind keeping it for a few more days? We might shortly have an acceptable fiber-optics streaming module for the D10." Of course I didn't have a D10; and had already written up the original then v2 iteration. "How about we send you a D10 with copper streaming module and an alternate fiber-optics module? This would be just to compare the two modules." I accepted so this story will continue in due course. Time to learn about optical LAN transmissions…