Sonoris’s mastering-oriented plug-in is designed to make dynamic control as unobtrusive as possible.
The term ‘mastering grade’ is often banded around in Internet discussions. There is a wide range of opinion, some of it informed, as to what kind of equipment the term applies to — though there is also some question as to whether or not the name actually makes any sense and properly applies to anything at all. At one end of this spectrum are the elitists, who admit only a shortlist of canonical equipment to the category; at the other are the pragmatists, who say “if it can be used for successful mastering, it is mastering-grade”.
Most of my mastering colleagues have a foot in both camps. There is a relatively small range of equipment, both digital and analogue, which by common consent is ‘for mastering’, but there are also certain pieces which are less generally useful in mastering, but which individual engineers swear by for particular projects. In the widest terms, however, ‘mastering grade’ is most often used to refer to processors which are clean, and which change the sound only in ways the engineer can completely control — perhaps by offering a particularly fine set of parameter increments and/or switched rather than continuous controls, for easy recall.
Like Sonoris’s Mastering EQ (reviewed in SOS March 2013: www.soundonsound.com/sos/mar13/articles/sonoris.htm), then, the company’s new Mastering Compressor plug-in has been designed by Peter Stenekes to be, above all, clean-sounding. It is, as a consequence, relatively sparse, with very few bells and almost no whistles at all. (All major plug-in formats are supported on Mac OS and Windows.)
The Sonoris Mastering Compressor has all of the standard features you would expect in a digital compressor, namely finely controllable Ratio, Threshold and Attack and Release parameters. But in addition to these it also boasts not just one, but two Knee controls, plus a very effective Auto Attack/Release setting, a side-chain filter which can be set to either high-pass, band-pass or low-pass, and a ‘harmonic enhancement’ feature for warming up slightly edgy material. There’s comprehensive control over the extent to which the stereo channels are linked, and the option to run it in Mid/Sides instead of left/right mode. You also get a choice between feed-forward and feedback compression paths, and the ability to feed a variable proportion of the dry signal back into the compressed signal at the output, either to help to restore transients or to enable those parallel or upward modes of compression which enjoyed some vogue a few years ago (but which I have to admit to not ever using much myself).
The dual-knee design is intended to ensure maximum transparency. The first knee, in a fairly standard manner, progressively softens the transition to compression as the signal begins to exceed the threshold. The second knee progressively reduces the ratio back towards 1:1 further above the threshold, in effect limiting the amount of compression that can take place. This, as we’ll see, is a very useful feature for music that has need for conventional compression control but which also contains the occasional legitimate ‘spike’ that needs to be let through.
As can be seen from the screen captures, the user interface is very plain, almost austere. The main part of the graphic is taken up with a large display which indicates the shape of the compression curve defined by the ratio, threshold knee controls, and below this are the controls themselves. The right-hand side of the graphic contains the gain-reduction meter, which can be switched to display a range of 6, 12 or 18 dB, plus a Settings button for saving and loading presets, a Monitor button which enables auditioning of the side-chain and the slider for the dry signal level.
During the time I had the Sonoris Mastering Compressor for review I was asked to master a recital of Vivaldi cantatas, performed by the Irish soprano Deirdre Moynihan, accompanied by a small string and harpsichord ensemble. It was a top-notch recording and a lovely performance, but the producer felt that it needed something of the ‘studio’ taken out (see box, right) to restore more of the smoothness and blend of a natural performance. The project took a lot of careful listening but, in the end, very little processing: just the Mastering Compressor at the front of the chain, and then my mighty analogue Knif duo (Soma EQ and VMII) doing their thing at the end of it. The SMCP was used to control those problematic aspects of the voice I have described, the EQ was to reduce some string resonances and add a slight whisper to the voice and some overall air, and the VMII was just idling, doing whatever it is that it does so very well.
In order to set up the Mastering Compressor for this task, though, I had to throw away just about all of my preconceptions about compression and most of my previous practice. The first deviation was the decision to use it in Mid/Sides mode — a compression configuration which I very rarely use, as it seems to me all too easy to mess with things that are not being listened for and then to discover them later. The second was to allow the compressor to make my decisions for me, by setting both the attack and the release parameters to programme-dependent automatic. I decided not to use the harmonic warming feature, and so the only variables I had to make any decisions about were the ratio (1.5:1 for the most part), the thresholds, and the side-chain filter frequencies.
Like many other mastering engineers, my work often begins with a simple psychological response; and when I experience this feeling of unease, I try to bring my theoretical knowledge to bear to identify and alleviate whatever it is that is making me feel uncomfortable. And so for this project I simply listened a few times: sometimes more or less casually, enjoying the singing and noting if and when my enjoyment was interrupted, at other times more critically, being more aware of my responses and noting them down.
In the end, the Mastering Compressor was used on just about half of the songs, in each case because there were places where dynamic artifacts were noticeable and correctable. For each of these tracks I first set the side-chain high-pass filter frequency to around 250Hz (high for an overall mastering side-chain, not so much for minimal compression being applied to the Mid portion) and then, as things progressed, changed it after auditioning the side-chain signal to optimise the focus on the vocal. Then I guessed how much over-reach there was on louder transients and beginnings of words, and dialled in a threshold that produced that amount of gain reduction, before listening and adjusting either way in case too much had been smoothed or not yet enough.
What was quite amazing was how little of the compression could ever be heard as ‘too much’, and just how forgiving the Mastering Compressor was when I wanted more. Having listened to what I thought was a minimally compressed song, I turned to the meters and was pretty stunned to see an entirely inaudible 3 or 4 dB of momentary gain reduction on one or two peaks. Even when visually clued in I couldn’t hear anything unnatural.
The finished CD has received some excellent reviews in the major classical press, and in particular has been praised for the amazing vocal tone and control which Deirdre shows in her performance. The Sonoris compressor, of course, didn’t create any of that control — but in the way I have explained (see box), it allowed her control to be more faithfully represented by alleviating the artifacts of the recording.
One of the most common, and most firmly entrenched ideas about the mastering process is that the engineer is able to add ‘punch’ to mixes using compression. Every mastering engineer I know has had the experience of being handed a fairly limp mix and being asked to work some kind of magic with it, after Web chat has led a client to believe that this is not only possible but a regular part of the mastering process. There is no doubt that a mastering engineer can and very often does make a mix sound more punchy, but he or she is not adding anything: the ingredients for the punch need to be there in the first place, and all the mastering engineer is doing is changing the envelope of some of those ingredients, minimising overhang in some cases, artificially accenting a transient in others, and in most cases matching the release of the major rhythmical parts to the overall pulse of the song. Rather less than ‘adding punch’, but then again, rather more. My normal ‘puncher’ is the Pendulum OCL2, which has served my needs for mastering projects in many genres including hard rock and dance music. The usual approach I take is to handle low-end dynamics with the Maselec MLA3’s low band, operating below 120Hz or so, with the OCL2 tackling the rest of the signal.
I have been working quite regularly over the last few years with the extraordinary bass player Milo Matthews, a virtuoso from Hawaii who uses his talent very musically (no self-indulgent high-neck noodling) to produce songs that are feel-good and funky. For Milo’s latest single I decided to try out the Sonoris Mastering Compressor in place of the OCL2. I generally set the OCL2 side-chain filter at 156Hz, so did the same with the plug-ins.
The track was really nicely mixed, with energetic bass and percussion, an explosive Spanish guitar break, a rap break, and a mixture of male and female vocals. All it needed was the fairly classic compressor combination of medium attack and fast release to accentuate the percussion and more clearly delineate the pulse, and as the MLA3 has switched settings from which I’d chosen a 200ms release, I did the same for the Mastering Compressor. With an attack time of 30-something milliseconds to let the transients through, and the parallel slider set to be adding just a scintilla of the dry signal, the effect was just what we needed. In technical terms, we had a little more snap and stinky whomp than before, and a greater overall liveliness: of course EQ played a large part in this too (EQ generally plays a much larger part in dynamics than it gets credit for) but the Sonoris compressor’s contribution was very impressive.
In my experience, controlled and (when needed) characterful compression comes mainly from the analogue realm; but there are times when character and control are at odds, and at those times, one needs a compressor which can exercise exceedingly fine control over programme material, without adding audible artifacts. These compressors tend to be in the digital realm, and although I have used other digital processors with some success, I don’t think that any of them gave quite the same transparency of performance as the Sonoris. Transparency is not always the primary desideratum — but when it is needed, I have yet to find anything else that fills the need quite so well.
The two main alternatives to the Sonoris Mastering Compressor in roughly the same price range would seem to be the Fabfilter Pro-C and DMG Audio’s Compassion.
There are two main kinds of recording and mastering situation. Sometimes, we’re trying to use the considerable resources of modern recording and mastering techniques to create a certain kind of sound and impression. At other times, we do our very best to make sure that artifacts introduced during the recording process are minimised and, if possible, eliminated. My examples of use in this review present these two ends of the spectrum.
There is nothing quite so pure, and quite so revealing of the artifacts of music production, as the solo human voice — especially, perhaps, the solo female classical vocal. Compression is generally used very sparsely and very carefully in classical recordings, as the paramount aim is to avoid those audible side-effects which even high-quality compressors can introduce. Paradoxically, however, the entirely artificial nature of the recording process sometimes means that compression is actually required to preserve the natural dynamics, especially when what is being recorded is basically a live recital.
This might sound self-contradictory, but in fact it’s quite straightforward. A classical singer performing on stage in an auditorium is able to sing quite naturally, with no self-conscious concerns about incidental dynamic changes except, of course, those that are specifically called for by the composer. Although in any concert performance there will be quite a range of this kind of dynamic variation, there is a natural smoothing-out effect from the acoustic space between the singer and her seated audience. Small variations in dynamics are simply ironed out by the distance. However, this natural smoothing effect of an acoustic space is entirely lost in a recording simply because of the proximity of the microphone: so is it is entirely possible, and in fact quite common, for a good recording of a beautifully controlled voice to present too much dynamic variation. This variation is only over a relatively small range, but that is accentuated by the microphone, which as well as being closer than the natural audience, does not respond like the human ear to dynamics.
Programme-dependent settings in compressors attempt to mimic the way our ears work. A simple example which shows that our perception of auditory events is naturally ‘programme-dependent’ is the sound of thunder: the first loud thunderclap of the storm, coming after complete silence, is perceived very differently from the second thunderclap that follows a minute or so later when the auditory memory, and its physiological effects, are still operating. Our reaction is de-sensitised by the initial loud noise which then lessens the perceived impact of the second, equally loud, thunderclap — though if there is a long gap between thunderclaps, that initial sensitivity can return.
The ear also has a natural sensitivity to the human voice in particular which enables us to hear nuances of psychological and emotional tone, and it is these that can get accentuated by the artificiality of the recording process, changing the emotional effect of the voice. In a ‘natural’ context, a sudden increase in loudness or harshness generally indicates irritation or anger, which might be quite the opposite of what needs to be conveyed in a love aria. Ideally then, these unintentional and artificially over-accentuated dynamic changes should be corrected for, but as these are in the main subliminal and unconscious responses, they are very hard to audition for in the mastering process!