If you want your mixes to go to 11, Waves' new mastering limiter might be the weapon you need.
Sometimes new technology gives us new ways to screw things up. Take, for example, the digital mastering processors that appeared in the '90s. In the right hands, they could add that last bit of sparkle and polish that would make a finished mix sound like the work of a professional. In the wrong hands, however, they could turn a bad mix into an unlistenable master, eliminating any trace of dynamic variation. And when record company executives got their heads round the idea that these devices could be used to make their CDs louder than the competition, even the best mastering engineers found themselves under pressure to deliver flatlined mixes.
In this quest for ever-louder CDs, two of the most popular weapons have been Waves' L1 and L2 peak limiters, which the company call 'Ultramaximizers'. L1 and L2 have two basic aims in life. The first is to increase the subjective loudness of incoming digital audio, without introducing clipping or other obvious distortion. The second is to dither the final output from your digital audio workstation to 16-bit for CD mastering (see box).
The genius of L1 and L2 lies in their ease of use. There are only two major controls: Threshold sets a level below 0dB to which the input signal is limited, and automatically applies the same amount of make-up gain. Output Ceiling then 'scales' the results so that the output signal never exceeds the Ceiling value. The net gain in loudness is thus equal to the difference between the Threshold and Out Ceiling settings. Waves say that with most material, it should be possible to achieve at least a net increase of at least 6dB before distortion and other side-effects become obvious. I've sometimes found this claim a bit optimistic, but even so, there's no denying that L1 and L2 can achieve some impressive results. Equally, there's no denying that they can be pushed much, much too hard.
The hardware L2 and the software L1 and L2 have been around for quite a while now, and have been facing increasingly sophisticated competition. Waves' answer is L3, a clever multi-band implementation of the same concept. Like almost all Waves processors, it's available both as a TDM plug-in for Pro Tools (in which case you need to run it on an Accel card) and a multi-format native plug-in, in each case supporting sample rates up to 96kHz.
In L3, the incoming audio is divided into five frequency bands. Linear-phase crossovers are used to do the splitting, so unless you actually apply limiting, the output signal is identical to the input signal, albeit delayed by the limiter's lookahead value of 80ms. The main controls are still called Out Ceiling and Threshold, and they still apply globally: whatever you do to the individual bands, their combined level is limited to the Threshold value, before being scaled to hit the Ceiling.
The clever part of the design lies in the relationship between the peak detection and the attenuation. Conventional multi-band limiters split the signal into separate frequency bands and then limit each band independently. Here, however, the level in all five frequency bands is summed using a patent-pending algorithm Waves call the Peak Limiting Mixer. If, at a given instant, this sum exceeds the Threshold value, L3 works out the amount of attenuation that is needed and intelligently distributes it across the different frequency bands. By default, the bands that are attenuated the most are those that contain the most energy, but it's also possible to instruct L3 to concentrate the gain reduction in frequency bands where you think it will be less noticeable. The result, at least in theory, is that you can apply more overall limiting with fewer audible consequences. Waves say that we can expect their PLM technology to be put to different uses in future products.
There are two versions: the simplified L3 Ultramaximizer, which is now included in both the Platinum and Diamond Bundles, and the full-blown L3 Multimaximizer, which is a Diamond-only affair. (Buying L3 separately gets you both versions.) Ultramaximizer won't frighten off anyone who's used to the simplicity of L1 and L2; apart from the Threshold and Out Ceiling sliders and the dither settings, the only user controls are a pop-up list of preset Profiles and a single Release control. Multimaximizer, by contrast, adds a graphical display and a set of controls for each of the five frequency bands, along with a pop-up list of Master Release characteristics.
Even by Waves' high standards, the ergonomics of Multimaximizer's interface are impressive. Despite the number of parameters on display and the unfamiliar nature of settings like Priority and Separation, it's easy to take everything in at a glance, and nearly all the parameter changes you might want to make can be achieved using a single mouse movement in the graphical display.
The display itself is a model of elegance and a very useful analytical tool for fine-tuning your settings. It is animated to show the gain reduction as it is applied across the frequency spectrum. An orange line follows the gain reduction in real time, but with most material, this moves too fast to give you a clear idea of what's going on, so Waves have added a 'smoothed trail'; this is a blue shaded area that follows the orange line, but decays much more slowly, so you can see where gain reduction has been applied. There's also a dark red peak hold line, which shows the maximum level of gain reduction that was applied at every point in the frequency spectrum.
Superimposed on these traces are visual representations of the band-specific controls. Each band features a coloured diamond which can be moved in both vertical and horizontal planes. Horizontal movement adjusts the crossover frequency between a band and its immediate neighbour, whilst vertical movement adjusts both the Priority and Gain of that band.
Both the Ultramaximizer and Multimaximizer versions of L3 include dithering features that are exactly the same as those in L2, except that the options are chosen from pop-up menus rather than by clicking to cycle through the choices. You can set L3's output resolution to 24, 22, 20, 18 or 16 bits, choose either no dithering or two types of Waves' proprietary IDR algorithm, and select the amount of noise-shaping that the dither noise should have. It's all exactly as straightforward as it sounds.
The concept of Priority is at the heart of L3's design, and it's these controls that allow you to shape the psychoacoustic element of the process. By increasing the Priority of a band above zero, you're telling L3 that the material in that frequency range is important, so when it encounters a signal that exceeds the Threshold, it will apply proportionately more gain reduction in the other bands and less in that band. Conversely, if you give a band a negative Priority value, L3 will take the opportunity to dump proportionately more of that band's content when it needs to achieve a certain overall gain reduction. The Gain controls are more conventional, doing exactly what you'd expect — increasing or decreasing the level of each frequency band in a fixed rather than a dynamic fashion, allowing the Multimaximizer version of L3 to act as a five-band, linear-phase equaliser as well as a peak limiter. Moving the coloured diamonds in the graphical display, or the double-arrow controls in the pane beneath it, adjusts the Gain and Priority together.
The Separation control allows you to vary the behaviour of the peak detection algorithm. At the default of 100, it operates as a fully multi-band device, where each band's peak-detection sensor receives only the signal from that band. As you lower the Separation, each band's detector begins to receive signal from the other bands as well, and at a value of zero, each band's detector receives the full-bandwidth signal. This makes L3 behave more like its single-band predecessors, although you can still adjust the Priority settings in order to distribute the gain reduction unequally across the different bands.
Each of the five frequency bands also has its own Release time parameter, which is a big help in achieving more transparent results. Typically, you might want a slower release time for bass frequencies to prevent distortion, with a tighter response higher up the spectrum to allow L3 to react more naturally to transients such as snare hits. You can manually set the release time for each band, but for most material, you're better off choosing one of the four programme-dependent settings from the Master Release pop-up. These all scale the release times across the frequency spectrum, usually with a slower release at the bass end, but also vary them to a certain extent in response to the programme material. Of these, ARC is similar to the auto-release algorithm used in L1 and L2; Warm is more programme-dependent at the lower end and stays closer to the manual settings in the higher bands, while Scaled does the opposite and Aggressive provides the tightest of the four release characteristics. The choice of Master Release algorithm makes more difference to the sound of the results than almost any other single parameter in L3. For rock and pop music, I usually preferred the ARC or Aggressive settings, while softer material often suited the Warm preset.
Another feature that's intended to help fine-tune the settings is the ability to solo individual bands. This allows you to check which instruments in your mix fall into which frequency bands, which can be handy.
Waves recommend the following minimum spec for any machine running their version 5 plug-ins (which includes both versions of L3):
- Mac: 867MHz G4 or better, 512MB RAM, Mac OS 10.2.8 or newer, 800 x 600-pixel display.
- Windows: 1GHz Pentium III, Pentium 4 or Athlon, 256MB RAM, Windows XP SP1 or 2000 SP4, 800 x 600-pixel display.
Whilst you can get away with running some of the simpler Waves plug-ins on lesser machines, L3 puts obvious demands on the CPU and I wouldn't want to run it on any computer that was showing its age.
Despite the welter of new controls, getting started with L3 is just as easy as it was with L1 and L2. You set the Out Ceiling control just below 0dB (allowing your signal to peak at 0dB risks the possibility of inter-sample peaks exceeding 0dB, which will clip when cheap D-A converters attempt to reconstitute your music) and bring down the Threshold control until you start to hear audible side-effects. At that point, you back off the Threshold a little, before getting into more detailed editing if you feel the need. One of the great things about all the L-series processors is that it's easy to A/B the processed and unprocessed signals at the same volume; you simply bring the Out Ceiling slider down to the same level as the Threshold slider, whereupon the perceived loudness should be the same as with L3 bypassed. (This would be even easier if Waves added a dedicated button.)
I suspect that many users will be perfectly happy with the simpler L3 Ultramaximizer, but those who take their mastering seriously will find the full Multimaximizer does things that can't easily achieve in other ways, and will be particularly useful in tackling problematic material. For instance, suppose you have a track that's based around a very solid bass sound, and you find that conventional wide-band limiting is causing the low end to pump in an undesirable way. This is no problem for Multimaximizer: simply set the range of the low or low-mid band to encompass the bass, and whack its Priority up. Now, when the peak detector encounters a big snare transient or guitar chord, it'll leave the bass alone and achieve the necessary limiting by cutting the other bands.
Likewise, wideband limiting sometimes results in a situation where slow release times leave obvious 'holes' after snare hits, but a faster release causes the bass to distort. L3's adaptive release algorithms allow you to use an aggressive fast release in the mid frequencies and be more gentle on the bass. L3's flexible design also allows you to process your mixes in ways that are more characterful than transparent. The available Profiles include settings such as 'Extreme Analog', which uses fixed rather than adaptive release times to create a deliberate pumping effect that could well suit some types of music.
In general, I think it's reasonable to say that compared to its predecessors, L3 will often give you either an extra couple of dBs of gain reduction for the same perceived level of distortion, or the same amount of gain reduction with somewhat fewer side-effects. However, the loudness race has got to a point where there really isn't all that much room left for manouvre. For example, where snare drums are prominent in a mix, I've always found that you can't push L1 or L2 all that far before these start to acquire a papery, thin quality, and that this tends to be the factor that limits the amount of the process you can apply. I was interested to see whether L3 would make it possible to achieve more gain reduction whilst retaining a solid snare sound, but I had only limited success. Turning up the Priority in the mid and high-mid bands did seem to give the snare more body, but only at the expense of obvious pumping in the bass and low-mid bands. I didn't find this any more pleasant than the papery snares I got with full-bandwidth limiting, but I suppose at least it's a different compromise, which might be more appropriate in some situations.
One of my other test tracks was a laid-back folk tune where the scope for wide-band limiting was restricted by the prominent female vocal, which had to be treated carefully to prevent it distorting very obviously on loud consonants. Again, dumping some of the gain reduction into the bass and low mids helped somewhat, but the results were still less than perfect.
I had a lot more success with rocky tunes containing plenty of distorted guitars, perhaps because these were masking some of the adverse effects. Even where the tracks were pretty heavily squashed already, L3 squeezed another 3 or 4 dB of gain out of the mix without making it noticeably worse, and some interesting pumping effects could be achieved. Where the test tracks featured contrasting sections, such as a sparse verse followed by a busier chorus, it was invariably helpful to use different settings in different areas, but then this is true of many mastering processes.
However much we might bemoan the self-defeating quest for ever-louder CDs, it's now a fact of life, and few of us can afford to put out discs that are 12dB quieter than everyone else's. In that context, Waves' L3 is an impressive piece of software design. It allows you to achieve a bigger boost in levels with fewer adverse side-effects, it's powerful yet easy to use, and the Multimaximizer version has the flexibility to tackle mixes that would be a headache for other processors. The lack of 192kHz support might be an obstacle for those working with high-resolution playback formats, but I would hope that audiophile material wouldn't be treated to this kind of processing in any case. Price-wise, it's far from an impulse buy, but compares well to rival products such as TC Electronic's MD3 Mastering Tools, and owners of the Platinum or Diamond Bundles who subscribe to Waves' Upgrade Plan should be able to get it at no extra cost. Now, if I can only squeeze another 2dB out of my track, it's sure to be number one by the time you read this...