SPL Phonitor

Whatever your stance on this last issue, there can be no doubt that headphones do have an important role to play in music production, and that a decent 'headphone plus amp' setup is a useful element of the engineer's toolkit. SPL are clearly aiming their new Phonitor at this demanding professional market.

What this means is that if we can learn the headphone system — and learn to make judgements relative to its non-linear characteristics (there's simply no such thing as a purely linear system: 'straight wire with gain' is a marketing term, not an engineering one) — then we should be able to make allowances for their particular distortions and produce high-quality work. Actually, I use headphones pretty regularly. For judging reverb tails in acoustic music, for example, I'll more than likely reach for a headphone setup, knowing that under normal listening conditions these tails will be significantly less audible. Nonetheless, I can temporarily tune out the actual amount of reverb to make a more accurate judgement of the other parameters (such as early reflections and decay) much more easily than on in-room monitors, and then tune back into the amount and compensate for the headphones' exaggeration of this.

Of course, headphones produce a particular kind of distortion of the sonic truth, and based on what I've said above, it wouldn't be unreasonable to assume that if there were a way of compensating for this at the source, we'd probably find fine judgements easier, simply because we'd need to make fewer compensatory adjustments to what we were hearing. This, it seems, is one of the main the motivations behind the Phonitor — combining a high-quality headphone amp with technology to overcome some of the problems of headphone monitoring.

Some of the switches are for the standard operations found on many professional monitoring units — such as solo, phase, dim (here a fixed 20dB), meter mode (VU or PPM), and signal-present and signal-overload indicators on the meters themselves. As the nicely written and informative manual proudly points out, though, the presence of further knobs for the Crossfeed, Speaker Angle, and Center Level functions indicate that "this is not just another headphone amplifier." As I suggested above, these functions are intended to provide the listener with a level of loudspeaker realism that's missing in standard headphone amps, in particular by providing the ability to better judge room ambience and spatial balance, which is allegedly all but impossible in standard headphone use, due to the artificial complete separation of the left and right channels. This spatial distortion can also affect our ability to hear a correct spectral balance: for example, a slightly muddy mix that's completely separated into two non-interacting channels can present with greater clarity over headphones, which can lead you to make poor judgements.

If what we heard through each ear was exactly the same, we'd be hearing mono. For those with a normal complement of ears, true mono can only be created by a mono source directly in front of a forward-facing listener and radiating equally to both ears. If we're listening through spaced stereo loudspeakers, we're not hearing true mono, even when we're listening to a mono source, because the shading effects of the head are still working to create an absolute difference at each ear, but being the same difference, as it were, they cancel each other out. Stereo effects, such as the positioning of phantom images, work by exploiting the differences between what we hear in either ear — differences that our brain then processes to produce a perceptual stereo 'field'.

There are two types of difference: timing (or phase) and amplitude (or intensity), and in general they work together. The reason why a sound source in front of you and to your left is aurally locatable at that spatial point is that the signal reaches your left ear earlier, and with greater amplitude, than it does your right. But although they generally work in tandem, the differences actually operate over different frequency ranges: below about 700Hz our ears respond strongly to the time differences, whereas above about 2kHz the difference in amplitude plays a greater role.

This is why different types of stereo microphone array in recordings give different results: coincident (X-Y) pairs pick up no timing differences, whereas spaced (A-B) pairs do — and this is partly why coincident arrays produce sharper stereo imaging and spaced pairs have a (phase-induced) 'bloom'. Near-coincident arrays, such as ORTF, combine a bit of both — which explains their popularity. If we now consider what's going on in headphone listening that's different from normal loudspeaker monitoring (one speaker per ear with zero interaction between them), we can immediately see what's missing from the headphone experience, and just what the three 'extra' Phonitor functions are intended to supply.

The first, the Speaker Angle control, "determines the stereo width caused by frequency-dependent interaural time differences" and is variable between 15 and 75 degrees, representing time delays of 90 to 635 microseconds. (Note that these angles are nominal approximations, for relative adjustment only.) The angle for ordinary headphone listening is assumed to be 180 degrees, so this reduction is supposed to bring the image more to the front.

The Crossfeed control "simulates the frequency-dependent interaural level differences", and can be set to one of six unquantified steps, from min to max. This parameter is not so obviously related to its real-world counterpart. Although the manual explains that "one can compare variations in crossfeed values as approximating the influence of different room sizes and characteristics on a given loudspeaker setup", I found this as about as useful as a Sarah Palin policy explanation — and simply followed the other advice given in the manual: "engage in your own additional fine-tuning." This is necessary anyway, as the two parameters interact, to a certain degree. When you've narrowed the image and crossfed the levels, it's likely that the centre of the image will be too intense, so there's a third control (Center Cut) that provides gain reduction of the mono sum of the left and right channels, in six steps between 0.3dB and 2dB.

At first, I used the Phonitor in standard headphone mode — reasoning that if this was unimpressive or inaccurate, no amount of electrical gerrymandering of the signal would be worthwhile. We now have quite a collection of high-quality headphone amps at URM, including a Benchmark DAC1, a Cranesong Avocet, a Graham Slee 'Solo' audiophile-grade headphone amp, a pair of Stax SR404 ear-speakers (driven by the SR600t drive unit), and — just sneaking in under the wire — the first production model of the DACS Headmaster, which combines the company's own headphone amp with the excellent D-A converter found in Crookwood's C2 monitor controller. For the comparative tests, I simply connected the Phonitor and the Slee to the balanced analogue outs of a Mark Levinson 39 CD processor (an ancient, but very fine, audiophile CD player) and the D-A equipped units (Benchmark, Avocet and Headmaster) to its digital outputs. I won't pretend that this was a scientific comparison, but it at least provided me with a sensible set of reference points.

The result — over a wide range of different music, from solar-powered Jack Johnson to Janaceck — was enjoyable and, literally, quite revealing. The SPL easily matched the best of the others (the Stax, the Slee, and the Headmaster) in terms of detail and accuracy of the frequency range. By comparison, the Avocet — an independently excellent monitor controller — sounded slightly flat, and the Benchmark sounded distinctly over-warm, and clouded over some of the details that the better units presented quite clearly. As these trials continued, I increasingly came to think that whatever overkill the 120V technology might be thought to be (the best audiophile units generally only have a third of that figure), it was certainly doing something good — and by the end of this section of the tests, I'd decided that the SPL was better, even, than the comparatively priced Stax, which had been my previous 'fatigue-free' classical reference of choice.

There are some very odd statements in the manual. For example, SPL say that "classical music stereophony recordings involve a much more restricted stereo width than pop or rock music." This is utter crap: not only does well-recorded classical music always have full-width imaging, it also has additional complex imagining within the L-R panorama and a dimensional depth that is rare in "pop or rock music". If anything, classical music is the perfect material for stereophony testing.

Still, I didn't want to invalidate my results by restricting my genres, so the first critical application for which I used the extra functions of Phonitor was during the mixing of some tracks featuring a large acoustic 'pop' group, comprising vocals, guitar, violin, and acoustic bass. The main tracks were all captured 'live', with the musicians sitting pretty much in a circle with a mic in front of each of them (think Buena Vista Social Club), but there were guitar, vocal and electric bass overdubs to mix in as well. There was a large amount of deliberate bleeding between the mics on the main tracks, and the mixing tasks were to create a realistic stereo image from the half a dozen or so mics, to make sure that in the mix the image I'd chosen didn't induce any sonic nasties, and to space and place the overdubs (with reverb and panning) to give them a natural fit with the other tracks.

Using the Phonitor for these tasks seemed to make things easier for me than I'd expected: the effect of the extra controls was not heavily pronounced (although I found it almost impossible to resist turning them on and then off — and then on, and then off again, to make sure that I was hearing a difference), but it appeared at the time that I could quite quickly dial in the elements of the stereo spread. I was convinced that this was better with the controls on than with them off. When I subsequently checked the mixes through the main monitors I was pleased with the result, but I was surprised to find that there were no surprises when I then re-checked the mix using my standard headphone set up, and going back to the Phonitor and turning the stereo controls on and off. The presentations were all different — as I'd expected they would be — but they also all sounded 'right', which was more of a surprise.

My second task was an unusual mixing session, in which I had to set up the pan positions of three stereo pairs of microphones that had been used to capture a string octet playing a heavily rhythmic piece of contemporary classical music. The octet (four violins, two violas, two cellos) had been seated in a fairly deep horse-shoe shape, and the positions of the mics captured a particular spread of sound at different 'depths' of the horse-shoe: a main pair of cardioids set up in ORTF; omni outriggers positioned slightly narrower than standard 'orchestral' spacing; and a closer 'spot' pair of sub-omnis (for the middle strings in the bend of the horseshoe). As before, I found myself quite quickly and confidently setting the pan positions — and, again as before, then turning the special circuits off and finding that it sounded different, but still 'right', on both settings.

The final tests were performed during the testing of Bob Katz's new 'stereoization' plug-in by Algorithmix. I was expecting that with the Phonitor circuits engaged I'd be able more accurately to judge the amount of processing to use — but with my previous results in mind, I decided to test things in reverse order. Wanting to add a bloom to a stark recording of guitar and vocal that had both been recorded in mono, I first set up the K-stereo parameters using the Stax headphones, and created a more realistic and natural image. Switching to the Phonitor in passive mode, I perhaps noted a very minimal enhancement of detail, but nothing that made me revise my initial settings. However, switching to active mode, there was a clear difference (obviously: the enhancement from the plug-in was purely to the stereo image), but still not in a way that made me want to make changes to the settings I'd made when using the Stax. I began to see something of a pattern emerging...

Listening to the decisions made using the new controls on a more normal headphone amp, I thought "yes, that's right"; rather than "oh, the non-Phonitor decisions were wrong". But although I wouldn't have made different judgements, the new controls perhaps made those judgements a little easier to make. It's a bit like using an FFT graphic to check your perception of tonal balance — you'd never mix from such a screen, but having mixed, it's good to see on them what you were expecting. So for those who aren't used to mixing on headphones I can see that a system like this might be helpful — although, having said that, you'll need to work just as hard to acclimatise to mixing on headphones as you would a new pair of monitors.

But the stereo trickery is only one aspect of the Phonitor, and I wonder if the meat of the matter is not really in the new controls themselves, but in the extraordinarily high quality of the headphone feed in which they are embedded. This might actually be good news for SPL, because although they seem to be pushing the new circuits as the unique selling point of this unit, there are actually a good number of competitor implementations of what's bascially the same theoretical idea (from RA Designs' free digital crossfeed plug-in to Ultrasone's physical placement of the diaphragms of their headphones). To my ears, it is the unquestionable quality of the normal mode of operation that's actually the more unique point!