This review page is supported in part by the sponsors whose ad banners are displayed below

Reality check. Expectation management. Process vs. results: Spatial's approach is clearly a fluidly moving target. As new software and firmware become available, upgrades or new features are just a download and perhaps subsequent recalibration away. Keeping website info current to reflect latest status will be important for Spatial to properly manage customer expectations.


About certain details in the calibration process, my expectations reflected Walter's initial verbal presentation and what Clayton's website had at the time. This put me somewhat at odds with 1st-round observations. I learned why only after the fact. Better communications would have avoided this.


Three months prior to delivery of their preloaded Mac Mini, Spatial had quietly shifted to Pure Music's core platform. Originally PM couldn't accommodate all of Spatial's prior features. Channel D's consumer interface had never been designed for this application after all. Because Clayton writes no software, it was left to Rob Robertson to integrate these features into his OEM version for Spatial HD.


To facilitate an already much delayed review, Clayton had simply opted to dispatch an earlier Spatial/Pure Music version, then failed to communicate the implications and what added functionality one might expect in a few months' time. Keep this in mind during what follows.


With the microphone set at one meter from the CS2.3 coaxial's center, Clayton ran more than an hour's worth of nearfield sweeps and corrections. The advertised mastering-quality studio reference EQ was Apple's freeware AU Graphic Equalizer in 31-band mode. While many customers likely won't sit through this calibration, I did for a good stretch. I had expected one initial measurement (or suite) input into a Trinnov-style program or ARC/Audyssey MultEQ plug-in to automatically calculate and apply all necessary corrections against a target curve predetermined by Clayton. After all, that's what massive modern computing power is good at.


Instead I watched something I thought outright primitive and on the face of it only crudely effective. Clayton manually manipulated virtual EQ sliders and compared each move against a new measurement. This went on for a good hour.
 
I observed how amongst many ragged response irregularities the uncorrected coaxial driver suffered a very serious suck-out at 2.000Hz and an equally large spike at 12.000Hz. Without digital repairs, this transducer would seem broken and unfit for hifi purposes even with its passive analog filter network. I also watched Clayton dumb down his measurement graphs with 1/3rd octave smoothing to have a better chance at making a mere 31 bands of manually operated equalization work. He also kept resetting the min/max graph extents of the x/y axes. This flattens out tall peaks and deep troughs by stretching the graph laterally and compressing it vertically. It reminded me why early on I had decided not to bother with review measurements. The sidebar explains what's involved. The upshot is, manual error correction against laterally stretched, vertically compressed maximally smoothed response graphs seemed quite primitive still. When Clayton signed off on his nearfield session, customers who had observed the process and final response prior to smoothing and re-scaling could have questioned whether leaving it there was really good enough. Those who expected close-tolerance matching of driver nonlinearities between left and right speaker would have been disappointed. Clayton only measured and corrected one speaker (the left). I'd been set up to expect more.


This continued with the far-field measurement session. Here the microphone was at ear height and in the place usually occupied by the listening seat. Clayton used the FuzzMeasure software for data acquisition but once again only ran the left speaker. This didn't account for dissimilar boundary positioning of the speakers or any room asymmetries and how they affect stereo response at the ear. Without having heard any music played by this system, I felt oversold and under-delivered observing the calibration. I was surprised too by the end result. The maximally smoothed/prettified final response looked reasonably flat from 20Hz to about 2.000Hz (having seen it at higher resolution I knew better of course) but thereafter rolled off to be down about 8dB at 10kHz. This treble response target seemed an experienced guess on Clayton's part after I'd told him that I sit about 5 meters from the speakers.


Once he remotely triggered playback of one of his preloaded tracks, I knew his calibration session was over. I hit pause and moments later heard the phone ring. Now I asked Clayton. Why had he only measured the left speaker's near-field response and only run the same speaker for the farfield response? "It's a start" and "I'm waiting for a plug-in that allows for stereo farfield testing" were the respective answers. If the HD in Spatial HD meant high definition, these seemed low standards. On why he relied on manual EQing over just 31 bands rather than far finer automated correction, Clayton explained that Spatial's previous iteration had included Audyssey-based functionality. He simply learned that automated equalization including DEQX was sonically far less transparent and also less flexible to manipulate. While functionally a huge step backwards, the slow manual—and dare I say far less exacting—31-band EQing I observed is claimed to be a significant forward step sonically.


Taking snap-shot notes of various EQ values he dialled in before closing the program some time later, I wrote down +7.83dB at 25Hz, +3.4dB at 30Hz, +3.2 at 40, +4.35 at 125, -2dB at 500, +3.7 at 1.000, -4.64 at 1600, +4.35 at 5.000, a massive +8dB at 8.000Hz and +4dB at 12.000Hz. I also saw that the digital crossover ended up at 160Hz with 4th-order 24dB/octave slopes for both hi/lo pass on the dipole 15-inch woofers. All this will be unique to each installation. What it does hint at are the degrees of compensation employed.
  
Sidebar. Unless one compares a maximally smoothed graph against its raw 1/48th-octave version and understands how manipulation of display scale affects the apparent magnitude of deviations, frequency response measurements are highly deceptive. Envision what happens when your 20 to 20.000Hz horizontal axis is stretched from 5 inches to 20 inches whilst the vertical amplitude axis in dB gets squashed from say 10 inches to 5. None of the values are tampered with. The deception occurs on visual impact. After also being maximally smoothed, the curve suddenly becomes reassuringly flat-ish. All of this easily hides ±10dB or larger deviations. It dumbs down displayed resolution and only shows very generalized trends, not the damning details one would want to correct in a Spatial HD context.
My crude hand-drawn example illustrates the point. I stretched line 2 horizontally and compressed it vertically. I then locked both dimensions again and scaled back width to match line 1. If these squiggles were accompanied by two matched scales—horizontally in Hz, vertically in dB—you'd see that both represent the same values. Line 2 simply looks a lot less alarming. Mind you, that's without any applied smoothing. If we applied max smoothing, the second line would look like a very shallow and smooth hill with just a few dB of elevation. That's the deception. It's no outright lie because translating the relation of vertical vs horizontal scale would tell us everything. If it occurred prior to smoothing. Post smoothing (i.e. what the usual review graphs are) the viewer has no idea just how many gross deviations from flat were hidden.


On a different note, consider dynamic range. Audiophiles get all moist over 130dB-range D/A chip specs. A low measured daytime background noise in an average dwelling hovers around 40dB above zero. A good quiet night-time equivalent might reduce that to 30dB. Most sane listeners find 95dB peaks in their listening seat loud enough. That makes for 55dB effective dynamic range during the day. To hear 100dB would need playback levels of 140dB peaks. 14-bit resolution or 16.384 discrete amplitude values in 0.005dB increments as Philips offered in their very first CD player suddenly seems quite sufficient on that count. Now add that most modern music is dynamically very compressed. Pure Music's display offers a dynamic range option which shows a dB figure to reflect the playing track's window between lowest and highest amplitude. On most music you'll be lucky to see 20dB.




Trying to access my iMac's iTunes library through the Mac Mini proved unsuccessful at this point despite what Charles had done. Clayton didn't know how to fix it and instead pinged Rob Robinson. Very shortly afterwards—on a Saturday no less, one week before Xmas—Robinson checked in by email. He advised that I install TeamViewer also on the iMac, then give him the respective session passwords for both Apple machines. This facilitated a remote trouble-shooting session to get the machines properly networked. The culprit turned out to be where my iTunes library was on the iMac. It sat in a private user account which the Mac Mini couldn't access. (This Channel D link provides valuable PC audio information. Read it before making related purchases.) Observing Rob reconfigure two Swiss computers from his US location whilst communicating with me in real time over a virtual note book was neat. His professionalism was absolute top drawer! To get access to his expertise (Channel D of course remains his primary focus) clearly was quite the coup for the Liederman/Shaw team. Here is a small writeup on Pure Music's functionality.

Intermission: Digital signal manipulation for room/speaker correction, volume, balance and certain effects via plug-ins like vacuum tube emulation is the future. It's intrinsic to working with digital files. I see two basic aspects: software + service. As software becomes more sophisticated to require greater computing expertise before it can be fully exploited, third-party service providers will likely interface with customers by remote access to perform on-site optimization and upgrades. An outfit like Spatial bundles pre-existing plug-ins with a third-party core platform—here a branded OEM version of Channel D's Pure Music—to act as service provider. This relies on properly trained personnel to interface with the end user. Software developer and service provider could be one and the same of course. In this open-source world, it matters not who wrote the code, who calibrates the system and who trouble-shoots it. In my case Charles secured initial remote access, Clayton performed calibration, Robinson added reconfiguration of iMac and Mac Mini.

What used to take dealer visits and hardware upgrades—the latter often costly by requiring you to send back your gear while facing down time, one-way ship fees and upgrade costs—can now occur in software. This has potential for attractive savings unless the new costs become consultation hours. Those should be built into a service contract. With Spatial it's built into the hardware package. It's reminiscent of custom install. That surged when home theater and whole-house automation were big. The main cost wasn't the hardware. It was installation labor. In Spatial's scheme it's customer service. With their present non-automated concept—check XTZ's collaboration with Dirac for a $450 self-amplified desktop speaker system that includes automated DSP correction—success relies on the remote consultant's expertise. It also relies on the consultancy pool he may cull from to resolve issues beyond his ken.

Spatial's approach is no longer entirely unique to suggest we're at the dawn of a new era. Consider Joe Cohen of the Lotus Group USA. He had dispatched this press release: "...The Lotus Group announces the G2 loudspeaker and Granada X1 active crossover module... each pair of G2 and Granada will be personally calibrated by our acoustical engineer to insure correct and precise alignment. The X1 will also be available as stand alone product working in conjunction with any loudspeaker's existing crossover to provide state-of-the-art room correction..."

Final corrected nearfield response on left speaker

Remote software configuration for intense digital signal manipulation is a new frontier. Eventually it should make the audio amplifier's output the only analog stage in the entire hifi chain. Spatial and competition presently find themselves simply at the very beginning of this. It's a steep learning curve (Clayton said he to date had configured some 50+ Spatial systems). Here learning is relative to available software; how best to package and streamline it whilst generating the most comprehensive measurements to apply effective room correction to the client's specific needs and preferences. It's a work in progress and based on my observations clearly so for Spatial as it should be for anyone in this sector where changes can occur from one month to the next.

Curious what plug-ins Spatial HD had preloaded, I came up with the following list of three sources - TC Electronik, mda-vst and Apple. TC provided Assimilator Native, Fabrik C/R Studio, M40 Reverb, Integrator and ResFilter Native. mda provided Combo, Dynamics, Talkbox, Testtone, Degrade, Leslie, Tracker, Ring Mod, Stereo Simulator, Shepard Tones, Re-Psycho!, Detune, Dub Delay, Sub Bass Syntheszier and Round Pan. Apple, all in their AU family of software, provided BandPass, DynamicsProcessor, Delay, Distortion, Filter, GraphicEQ, HiPass, HighShelfFilter, PeakLimiter, LowPass, LowShelfFilter, MultiBandCompressor, MatrixReverb, ParametricEQ, RogerBeep, SampleDelay and Pitch.

A good question to ask in all this is just how flawed hardware ingredients can be to ultimately not matter once digital corrections have 'fixed' them. Without digital EQ for example, Emerald Physics' raw speaker ingredients flatly wouldn't work, period. How important will loudspeaker design and parts quality be if lesser speakers can be massaged digitally to perform better? Would treating my €18.000 ASI Tango R speakers to the same digital room correction provide them with a performance advantage commensurate with their far higher price? Or would minimized room colorations level the playing field down to fixed items like dispersion pattern continuity and fully digital crossovers vs. digitally corrected analog crossovers?


Another very relevant question in any DSP scheme is management of gain. With Spatial HD's TC Konnekt package, the processor's digitally actuated analog volume control can't work in Spatial's preferred 24/192 mode. For that one must control system volume exclusively in Pure Music's interface. That means 64-bit digital volume*. For analog gain trim the TC must run in 24/96 or lower mode. This reactivates its front-panel control. Doing that can invite mishap however. That's because Pure Music is an automatic sample-rate switcher. As soon as a 16/44.1 track is followed by a 24/192 hi-rez file, the program switches to native 192 resolution. This defeats the analog volume control to deliver full-blast signal over 97dB speakers which—I can assure you—is bloody loud.


Thankfully I happened to be at the controls when this occurred the first and only time. I managed to cut signal over a few seconds. Those who mistrust deep or any digital attenuation (this system's abnormal combination of high speaker sensitivity, powerful amps and steep overall gain had me routinely invoke minus 55dB!) will want to limit the TC Konnekt to 24/96 mode and control volume in the analog domain. Just don't play 24/192 files. In my 5m x 12m room attached to an open second storey, I reached stout room levels over a very narrow range already, i.e. 8:00 on the analog control was plenty loud, 7:00 still equalled mute. A combination of limited digital signal trim plus analog volume would then be the ticket.


Let's also remember that Clayton performs gain matching between compression-loaded tweeter and big midrange as well as bi-amp balance and all compensation EQing in the digital domain. Even analog volume with the TC Konnekt won't ever bypass the considerable digital gain trim that's irrevocably built into this system to even work properly. To purists this curious fact alone could be fatal and negate the entire concept.
__________________________

* When I asked him about this specifically, Walter claimed perfect transparency for the 64-bit scheme at all levels. Rob Robertson meanwhile confirmed my suspicions. "I'll simply tell you what I tell all my Pure Music customers. For critical listening purposes, I consider acceptable a maximum of -12dB digital attenuation, preferably no more than -6dB." This fully jived with Daniel Weiss' provision of four resistive (analog) trim levels in my DAC2. This can match its output voltage to overall system gain so that digital attenuation for track-to-track and disc-to-disc level matching can remain as modest as possible.

Today's report barely skims the surface of this exciting topic to actually review this system. In fairness to Spatial's keep-it-simple concept whereby the customer is presented with a turnkey system and not encouraged to watch and second-guess the process, the above comments likely won't mirror buyer reactions. Buyers ought to be results oriented. How does it perform? Clayton also advised that during Q1 of 2011 Spatial HD would have different EQ software, facilities for stereo room measurements and more. A website update was imminent to reflect current facts. This rendered certain of my descriptions passé already. I chose to leave them part of the narrative simply to underscore how moving of a target Spatial's approach is by design. Fixed hardware is what it is. Software-based hifi becomes endlessly scaleable. It potentially gets better and better as technology advances and our understanding on how best to harness it grows from experience.

Enlarge!