Innuos Statement music server John Atkinson May 2020

John Atkinson wrote about the Innuos Statement in May 2020 (Vol.43 No.5):

When Jason Victor Serinus reviewed the expensive Statement server from Portuguese company Innuos (footnote 1) in the April 2020 issue of Stereophile, he concluded that "Innuos has created a transparent instrument that scores big in soundstage size and depth, dynamics, and bass reach." Comparing the Statement and a sample of the more affordable Roon Nucleus + (footnote 2) (fed by a linear power supply)—both servers feeding his dCS Rossini and EMM DV2 D/A processors via USB—he found that via the Statement "the treble seemed slightly rounded, the presentation a touch warmer than through the Nucleus +. . . . The Statement warmed the piano and smoothed out the top in a manner that some would call analoglike or tubelike."

These descriptions of sound quality suggested that I should be able to find measurable differences between the Statement and other sources of USB data. In addition, the review of the Statement in our sister magazine Hi-Fi News indicated substantially reduced jitter in some of the DACs that were connected to it. When the Statement arrived in my test lab, I didn't have Jason's dCS and EMM processors to hand, but I did have a Mytek Brooklyn and two AudioQuest DragonFlies, a Red and a Cobalt, that I could use with the Innuos server. Using my Audio Precision SYS2722 analyzer, I compared the spectra of each DAC's analog output signal receiving 16- and 24-bit J-Test data sampled at 44.1kHz via USB, either from my MacBook Pro laptop or from the Statement. To sum up my findings: While there were no measurable differences in the Mytek's output when fed data from the laptop or the Innuos server, I did find, with the AudioQuest DACs, that sourcing data from the Statement gave slightly cleaner spectra. The differences were at a very low level, however.

After the April issue was sent to the printers, I performed another series of tests with my long-term reference DAC, the PS Audio DirectStream, which is fitted with a network bridge card. (I am still using the "Snowmass" operating system, having not yet upgraded to the latest "Windom" firmware.) I used the Roon app to send the PS Audio J-Test data via my network and via USB from the sample of the Nucleus + that I had purchased following my review. (The Nucleus + was powered by its standard switching supply.)

Using the Innuos-recommended iPeng 9 player app on my iPad Mini, I then sent the J-Test data that I had copied to the Statement's internal storage to the PS Audio via USB. The Statement was running the same v1.4.3 operating system as it had when Jason auditioned it. (The iPeng settings screen revealed that this appeared to be a version of Logitech's Squeezelite firmware.) The Statement's volume control in the iPeng playback screen was set to its maximum for these tests, as was the PS Audio's.

Fig.1 PS Audio DirectStream, high-resolution jitter spectrum of analog output signal, 11.025kHz at –6dBFS, sampled at 44.1kHz with LSB toggled at 229Hz: 16-bit USB data sourced from Innuos Statement (left channel red). Center frequency of trace, 11.025kHz; frequency range, ±2kHz.

I found almost no differences in the DirectStream's analog output whether it received data from the Nucleus + or Statement via USB or from the Nucleus + via Ethernet. In all three cases, the odd-order harmonics of the low-frequency, LSB-level tone were very close to the correct levels. Compared with sending 16-bit J-Test data to the PS Audio DAC via an AES/ EBU link, both the USB and Ethernet connections resulted in an additional pair of sidebands at ±50Hz to the sides of the spectral spike that represents the high-level tone at exactly one-quarter the sample rate. These sidebands were slightly higher in level with network data than they were with USB data, but the levels were identical with USB data sourced from both the Nucleus + and the Statement (fig.1; note expanded vertical and horizontal scales).

Listening to the Servers
When I discussed my original findings with Stereophile Editor Jim Austin, he suggested I do some listening tests, comparing the Statement with my Nucleus +. Even though I am more of a Leveler compared with Sharpener Jason (footnote 3), I was intrigued by what I had found in the measurements and agreed. The system would be my KEF LS50 loudspeakers driven by Lamm M1.2 Reference monoblocks, these connected directly to the PS Audio DAC. AC was supplied from an AudioQuest Niagara 5000 Low-Z Power/Noise-Dissipation System.

The listening tests were divided into three kinds. First, using the Settings window on the local webpage, I set the Statement to be a Roon player. I then enabled the PS Audio DAC, connected to the Statement's USB output with a long Belkin Gold cable, as a Roon endpoint. (This was the only USB cable I had that could reach the PS Audio from the rack where I had the servers.) By using the Roon app to send the same data from the Nucleus + to the Statement or via AudioQuest Diamond Cat-7 Ethernet cables to the PS Audio's network bridge via my rather old NetGear router, I could compare the two sources by switching between "USB" and "Bridge" with the PS Audio's remote control. (I found that I had to set the Statement's volume in Roon to the maximum in order that the PS Audio's output level would be the same with both connections.) The comparisons were a little more complicated than I expected because Roon halted network playback when I changed the PS Audio's input to USB, even when I grouped the two zones. This comparison would also be between the Ethernet and USB connections, but I felt the ability to A/B switch would be a good starting point.

Fig.2 When you right-click on an album on the Statement's Library page at, it displays the track with its metadata, including format, sample rate, and, with a bit depth greater than 16, "HD."

For the first set of comparisons, I cued up "Flamenco Sketches" from Miles Davis's classic album Kind of Blue (24/192 AIFF file, Legacy/HDTracks). After an hour's repeated listening to this nine-minute track—levels were matched, of course—I felt that perhaps Jimmy Cobb's brushed snare-drum pattern was a tad better resolved with the Statement. Perhaps. Perhaps his softly struck cymbals had a touch more leading-edge definition with the Nucleus +. Perhaps. Paul Chambers's double-stopped notes on his bass were perhaps slightly better differentiated and a touch weightier with the Statement. Perhaps. The shimmering of the piano's sound due to the not-quite-in-tune strings in the notes with which Bill Evans opens his solo were equally clear with both servers.

Perhaps some orchestral music would be more revealing of differences? I played Simon Rattle and the Berlin Philharmonic's performance of Beethoven's Symphony No.5 (24/192 WAV, Berliner Philharmoniker BPHR 160091). Both servers sounded forceful on the familiar dotted-note motif in the first movement, with excellent weight to double basses. Both presented a good sense of space around solo instruments and a coherent image of the orchestra. Both allowed me to be caught up in the music making.

It was only with "It's Like That" from Us3's Hands on the Torch (16/44.1 FLAC, Blue Note) that I finally felt confident of hearing a difference between the two sources. With this hip-hop/jazz mashup, the bass line had a touch more drive with Statement sending data to the PS Audio, the Ethernet connection from the Nucleus + a touch more upper-bass bloom—but in both instances, just a touch.

Time for the second and perhaps more meaningful round of listening tests. I disabled the Statement as a Roon player so that I could compare the sound with data stored on the Statement's internal drive with the same data stored on the Nucleus +. By controlling the Nucleus + with the Roon app and the Statement with the iPeng app, I could again A/B the two sources.

Given that I was confident about hearing a difference with the Us3 track, I started my comparisons with it. Again, I could hear differences, and again they were in the lower-frequency region. The upper-bass bloom from the Nucleus + was evident, but while the low frequencies still seemed a little better defined with the data sourced from the Statement, the bass riff now seemed to lack a little mid-bass force. A little. Higher in frequency, the vocals sounded identical in both tonality and spatial presentation, as did the tunnel of reverb behind the sampled mono drums that punctuate the track.

To focus on the low-frequency differences, I followed "It's Like That" with Daft Punk's "Get Lucky" (24/88.2 FLAC file from Random Access Memories, Columbia/HDTracks), which I had copied to the Statement's library (fig.2). Nathan East's driving bass line had slightly better articulation played from the Statement, slightly more upper-bass bloom from the Nucleus +.

Before I got totally sick of "Get Lucky," I did one more set of comparisons, this time in turn connecting both servers' USB outputs to the PS Audio DAC using the same Belkin Gold cable. No doubt about it. The Roon server's USB output doesn't have quite the low-frequency clarity that its Ethernet connection does. (In a discussion I had some years back with the late Charley Hansen of Ayre, he was insistent that all things being equal, USB would sound better than Ethernet.) When it came to the USB connection, the Nucleus's bass was outclassed by the Statement. Not by much, I admit, but enough to matter. But I still didn't feel the Statement had a treble that was, in Jason's words, "smoothed out . . . in a manner that some would call analoglike or tubelike."

I cued up Patricia Barber's "The Beat Goes On," from Companion (DSD64 file, Mobile Fidelity Sound Lab), which I copied to the Statement's library. Yes! Both Barber's vocals and her Hammond organ solo did indeed sound slightly smoother, less incisive on the Innuos server. Score one for Jason.

For the final round of tests, I eliminated potential differences due to the USB and Ethernet cables by plugging an AudioQuest DragonFly Cobalt DAC into the USB ports of the Nucleus + and Statement. One of the tracks Jason had mentioned in his review was "Electrified II" from Yello's Toy (24/48 FLAC, Polydor/HDtracks). He described this track as being "a bit too hot and sizzly" and felt it was less fatiguing when played back from the Statement. Yes, it did sound "a bit too hot" played back on the Roon server with AudioQuest NightHawk headphones, though the bass was weighty enough. As with the Patricia Barber track, the treble was a touch smoother with the DragonFly plugged into the Innuos.

I was also impressed by the slightly better low-frequency control the DragonFly exhibited with the Statement.

Summing Up the Servers
So what to conclude from the two solid days I spent listening with two D/A processors to these two servers? Yes, there were differences, but even with the USB connections the differences were small. If I left the room then came back again, I don't think I could tell you whether the PS Audio DAC was receiving data via USB from the Statement or via Ethernet or USB from the Nucleus +.

Why did I find the differences between the Innuos and Roon servers more difficult to hear than Jason had? First, as I wrote many hundreds of words back, I tend to be a Leveler whereas Jason is a Sharpener. Second, I use its standard switching supply with my Nucleus +, whereas JVS powers his with a linear supply from Small Green Computer. And third, it is entirely possible that my KEF/Lamm/PS Audio system, as musically satisfying as I find it, is just not as resolving as Jason's Wilson/Dan D'Agostino/dCS system.

When I visited Jason's old place in Oakland, California, in 2012, his listening room/system was one of the finest I had experienced. I visited Jason's new home in the Pacific Northwest while preparing this follow-up review and found his current system very revealing. For example, the improvement that resulted from replacing the standard switch-mode power supply for his Roon Nucleus + with a linear supply was more audible than I anticipated. Also relevant is that Jason listens at SPLs that are at least 10dB higher than is possible with my system.

To sum up my experience of the Innuos Statement, this well-engineered server gave me hours of musical pleasure once the formal listening tests were finished—though I won't be playing those tracks from Kind of Blue, Hands on the Torch, Random Access Memories, Toy, or Companion any time soon!—John Atkinson

Footnote 1: The Innuos Statement, as reviewed, with a 2TB internal drive, costs $13,750.

Footnote 2: The Roon Nucleus + costs $2498 without any audio media storage. I added a Seagate 2TB BarraCuda 7mm hard drive, which cost $82. Roon Labs:

Footnote 3: See Robert Deutsch's 2009 essay on Sharpeners and Levelers here. Also see the introduction to Jason's review of the Wolf Alpha 3 SX elsewhere in this issue.


Bogolu Haranath's picture

Waiting for a review of the Innuos Phoenix USB Reclocker by JVS :-) ..........

chuckles304's picture

Here's to hoping no one accuses JVS of trying to spread plague by having his sister-in-law over or "waving his middle finger" at those of us who can't afford the server.

Other than that, nice review.

Jason Victor Serinus's picture

I'm glad you like the review.

My sister-in-law returned to Oakland safely on January 3. The only plague she carried with her was the hi-end audio bug.

Unfortunately, she also returned to her job teaching 4th graders. That's where she may have contracted COVID-19. She and her hubby have now fully recovered. Fingers are currently crossed for my neighbors two houses down. We only have 19 confirmed cases in the entire county, and three are my neighbors.

Stay safe.

Bogolu Haranath's picture

Uncle Sam may start tracking your neighborhood with GPS :-) ........

John Atkinson's picture
I have written a follow-up review of the Innuos server for the May issue, comparing it with the Roon Nucleus+.

John Atkinson
Technical Editor, Stereophile

Bogolu Haranath's picture

JA1 could also review the Innuos Phoenix USB Reclocker :-) ........

Bogolu Haranath's picture

May be JA1 could also review the new Aurender flag-ship W20SE music server :-) .......

DH's picture

The measurements are at a level of -130 to -145db; so any noise or differences are inaudible in any case, aren't they?
Would seem to belie the descriptions in the review.

In addition, the results with the Mytek would indicate that with any well engineered audiophile DAC the Statement makes no difference. The Dragonfly DACs aren't really audiophile, and the Cobalt actually measures worse than it's $100 cheaper brother.

thyname's picture

JVS: you have the Roon facts a bit wrong here. The experimental feature is when using Statement as Roon Core with SqueezeLite Player. I wish I could post a screen grab here to show you that setting on my Statement. There is nothing experimental here when using Statement as “regular” Roon Core only, or Roon Core + Player.

Here is the description on the experimental feature:

****(Experimental) New option in Roon Settings to allow using our internal player with Roon, enabling RAM playback. Please see setup instructions below.***

Also, I was shocked to hear you liked Nucleus more than Statement as a Roon Core. To me, Statement when used as Core + Roon Endpoint is significantly better than when used as a Player only with Roon Nucleus as the Core. Even the Zenith MK3 I used to own before the Statement is better that way. I suggest you give it another shot, and ask Innuos folks for help with the set up, as something must have been terribly wrong with the way you had configured stuff.

Habanero Monk's picture

My endpoint system buffers entire tracks. How does this server help vs my QNAP?

CG's picture

Perhaps the measurements would be more revealing if they were made using the Max Hold function in the analyzer. That would capture peaks in the noise spectra over a period of time. Averaging tends to, well, average incoherent or faux-coherent disturbances into almost nothingness. That's the reason for using averaging.

Many sources of noise and interference only appear every few samples. So, they'd get averaged out or only appear as a modest bump in the noise floor. In the communications system biz, engineers figured this out some time back. You can lose a couple symbols out of a thousand, and it only shows as a couple hundredths of a dB in averaged degradation. Yet, those bits are entirely corrupted.

People tend to listen to a continuum of samples - how silly! - which means that averaging isn't representative of how something might sound. A noise burst that only appears once every hundred samples might be pretty loud for those samples, but the average would only show a slight perturbation.

I'm not sure whether the most recent Audio Precision products have this function. Older units certainly did, and AP is not a company that skimps on useful design features.

This might be a trip down the rabbit hole in regard to audio. It only takes a couple mouse clicks and a small amount of time to see if there's anything there. (Easy for me to say - I'm not the guy doing the work!)

John Atkinson's picture
CG wrote:
Perhaps the measurements would be more revealing if they were made using the Max Hold function in the analyzer. That would capture peaks in the noise spectra over a period of time.

I will try this. I have been using averaging because so many of these things at very low levels become impossible to see without the noisefloor being lowered by the averaging.

CG wrote:
This might be a trip down the rabbit hole in regard to audio.

There's always that possibility, of course. :-)

John Atkinson
Technical Editor, Stereophile

CG's picture

This might be compared to (maybe) not seeing the forest for the trees.

But, you're partially retired, right? Mostly confined to home for now, right? No concerts or time at the pub, right? So, you probably are just looking for things to keep you busy.

No need to thank me...

Bogolu Haranath's picture

JA1 is currently busy, binge-watching movies on Netflix ......... He could take a break and do something useful in audio ...... Just kidding JA1 :-) ........

John Atkinson's picture
Bogolu Haranath wrote:
JA1 is currently busy, binge-watching movies on Netflix ......... He could take a break and do something useful in audio ...

I just posted my follow-up review of the Innuos Statement from the new (May) issue. It's appended to the end of Jason's review:

John Atkinson
Technical Editor, Stereophile

Bogolu Haranath's picture

I'm glad JA1 is binge-reviewing audio equipment instead of binge-watching movies :-) .......

Bogolu Haranath's picture

With all due respect ...... it is not a fair comparison ....... You (JA1) could have used at least Vimberg Mino speakers for Innuos comparison test, not LS-50 :-) .......

John Atkinson's picture
Bogolu Haranath wrote:
With all due respect ...... it is not a fair comparison ....... You (JA1) could have used at least Vimberg Mino speakers for Innuos comparison test, not LS-50 :-)

Perhaps the Mino speakers would have been more revealing. Perhaps. However, I wanted to use my regular reference system for the comparisons with the Nucleus +. More of a real-world context, if you wish.

John Atkinson
Technical Editor, Stereophile

direstraitsfan98's picture

The best streamers seem to jump in price from entry level ones.
I just want something with good features, an attractive looking chassis, and a price tag that makes sense for me. Doesn't seem like one exists that checks all these boxes... perhaps I will just make do running my computer into my DAC until I'm close to retirement age...

Bogolu Haranath's picture

Mytek Brooklyn Bridge (DAC) reviewed by Stereophile, is not very expensive ....... Of course, it also has a DAC :-) ........

direstraitsfan98's picture

After doing some research I found that Innous sells a much simpler and 'standard' version of their statement Statement model that as far as I can tell has the same features. The price? 10 percent of the statement's cost. I think I will look to get one of those. I'm not considering the Mytek because I have a dedicated DAC already.

Is it safe to say that the standard Innous Zen will offer up a large portion of the Innous Statement's performance, at a fraction of the cost? Probably.

Bogolu Haranath's picture

May I ask, what DAC you are using? ........ Does it offer several digital reconstruction filters like Mytek? :-) ........

Bogolu Haranath's picture

The new Innuos Zen models also offer 1 TB HDD storage :-) ........

Jason Victor Serinus's picture

I chose the Statement for review because when I heard it at the audio show, it was noticeably more transparent than the server beneath it, even with the USB reclocker added. Have you considered the Roon Nucleus or Nucleus + together with an outboard LPS? You have to use Roon, of course, which costs $, and there's no storage unless you add some. But it sure sounds great.

Bogolu Haranath's picture

When we add storage, Nucleus becomes an 'Atomic nucleus' :-) ........

Bogolu Haranath's picture

The only thing MBB can't do is, you can't use it to post comments on Stereophile website :-) .......

jameslockie's picture

I am currently using the supplied wall wart.
Be interested to learn of your experiences with External LPS for this device.

Jason Victor Serinus's picture

I discuss this in a review recently posted to AudioStream: Much smoother sound, greater detail, more saturated colors, far deeper bass... fill in the blanks.

In virtually every interview I've conducted with a manufacturer or designer, they have emphasized the supreme importance of the power supply or transformer. In my admittedly limited experience, what is true for internal power supplies applies equally to external power supplies. Unless a designer has made a superhuman effort to isolate the noise from a switch mode power supply, from my admittedly limited experience, a linear power supply will improve the sound. I'm not an electrical engineer, and cannot discuss this with more sophistication - for all I know, there's an isolated example where what I've said isn't true - but I've certainly heard the difference in my system. I would never want to go back.

There are undoubtedly better LPSs on the market than the ones I've tried, but I haven't had the financial resources or time to explore them. The ones I've used have made such a significant difference that I've been happy. Once we get past COVID-19, or I win the lottery...

jameslockie's picture

Thanks JVS.
Good article.
I will investigate LPS alternatives.

tresaino's picture

I don’t have the Statement but am very very happy with its smaller brother Zen Mk3. Software being the same, I never encountered the software problems JVS described. And sonically the Zen Mk3 delivers in my system, in spades. I don’t say this easily but over the years I learned reading between the lines and found Serinus’ review overall wrong and also unfair. I had a few servers and streamers before buying the Mk3 and never looked back. The Zen doesn’t cost a fortune, sounds much better and the software is very intuitive and also very reliable. I was almost upset reading JVS’ review. Thank to John Atkinson for a more balanced follow up review. Greetings from a Stereophile subscriber since more than 25 years.