You are here

Waves L2 Ultramaximizer

Digital Mastering Limiter By Paul White
Published December 2000

Waves L2 Ultramaximizer

Paul White lives life to the max with the new Waves L2, a mastering limiter with all the trimmings.

The original Waves L1 Ultramaximizer plug‑in for Sound Designer II has been a mainstay of my digital‑editing and mastering system for a number of years, yet its ability to coax increased subjective level out of an apparently 'maxed out' track never ceases to amaze me. Essentially, the L1 Ultramaximizer was a very clever look‑ahead limiter that managed to skim the top few dBs from the peaks in your mix without introducing audible side effects. However, it was also a masterpiece of ergonomic design — the usual rash of controls was replaced by just two main sliders, one setting the peak level at which you wanted your signal to end up, and the other increasing the input gain of the signal to push it up against this limiting threshold. A third control varied the limiter's release time. The result of this system was that you could both limit and normalise in one operation.

Included in the L1 was Waves' own IDR noise‑shaped dithering system developed in association with the late Michael Gerzon, a much‑respected and highly talented audio pioneer. IDR allows high bit‑depth signals to be reduced to any target bit depth while retaining as much dynamic range as possible — a number of noise‑shaping options are available to suit different types of material. Typically, the perceived dynamic range can be improved by up to 18dB over what would normally be expected for undithered signals of a given bit depth — which is equivalent to three additional bits of resolution! Two IDR options are available: Type 1 produces the best distortion and linearity figures while Type 2 is optimised for the lowest noise. There are also three 'depths' of IDR that can be used to provide perceived resolution improvements of 6, 12 and 18dB respectively — this is equivalent to adding dither noise to the value of one, two or three bits. Strictly speaking, you should always have the IDR switched on, both when reducing bit depth for any reason, and as the last process in the production chain — usually the maximum depth of IDR would only be used at the final stage of mastering.

Soft Habit To Break

The L2 offers impressive connectivity — AES‑EBU, S/PDIF, word clock, nine‑pin serial port, and both jacks and XLRs.The L2 offers impressive connectivity — AES‑EBU, S/PDIF, word clock, nine‑pin serial port, and both jacks and XLRs.

The L2 is basically a stand‑alone, 2U‑rack, hardware version of the plug‑in, which not only mimics the software functions, but which also has some extra refinements and comes with a host of professional interfacing for use in mastering systems. Stereo digital audio can be supported at any of the common bit depths and sample rates up to 24‑bit/96kHz, and can be received and passed on through either AES‑EBU or S/PDIF input and output sockets. A word clock input is provided, for synchronizing the unit to a master clock, and there is also a nine‑pin serial port on the rear panel, allowing for remote function control. Should you wish to feed analogue signals into the unit, both balanced and unbalanced I/O (on quarter‑inch jacks and XLRs) are catered for, operating at +4dBu and incorporating high‑quality 24‑bit converters.

Because the L2 Ultramaximizer can accept signals of up to 24‑bit resolution, and as it will usually be required to increase the level of these signals, the internal processing is performed to 48‑bit precision. This high resolution helps minimise the errors that inevitably occur when any digital data is changed in level (due to mathematical truncation errors). The fact that upward level changes are involved also means that there may be a benefit in using the noise‑shaped dither even when the input and output are both set at 16‑bit resolution, as the processed signal (inside the box) will almost certainly end up at a higher resolution than this.

The left‑hand side of the front panel is given over to a row of switches used to select the input type, clock source, sample rate, output bit‑depth and noise‑shaping options — the output resolution can be set to 16, 18, 20, 22 or 24 bits. Each channel has an input level trim followed by three larger controls, with associated three‑digit plasma displays and 16‑segment bar‑graph meters, that duplicate the functions of the software plug‑in. The Out Ceiling control determines the channel's desired peak output level, and the adjacent meter shows the output level. A peak‑hold facility is provided, and a separate button resets the peak hold for both channels.

The control labelled Threshold is that which raises the level of the input signal and pushes it against the limiter. The associated bar‑graph meter shows the level of the input signal. The final control sets the limiter's release time, with the meter above functioning as a high‑resolution gain‑reduction meter and allowing you to see exactly when limiting is taking place, and how much level reduction is being applied. Because a fixed release time doesn't always suit material that's constantly varying in dynamics, there's also an ARC Auto Release button that makes the release time programme‑dependent. In most cases, leaving ARC active gives the best results, as the system analyses both the peak and RMS characteristics of the input signal to predict the best release time. In general, transients are treated to a short release time while signals with a high RMS level are given a longer release time.

Finally, a stereo link button allows both channels to be locked for stereo operation, in which case everything is controlled from the uppermost channel. Though the two channels can be used independently to process two mono signals, there is only a single Bypass button for them both.

Two's Company

The original L1 software was always easy to use and it invariably produced first‑class results. Happily, the L2 Ultramaximizer is just as easy to use and has clearer metering. Setting up is about as simple as you could hope for — you decide how loud you want your peaks to be in relation to digital full scale (a dB or two below maximum is a good idea), and then adjust the Threshold control until the gain‑reduction meter shows that the loudest peaks are being limited. Some skill is necessary in determining just how much limiting you can apply before the signal starts to sound unnatural, but that's about the extent of it. In most cases, you can apply at least 4 to 6dB of peak limiting, and often more, to an already normalized signal without hearing any side effects. That's equivalent to a doubling of the signal level!

Using the L2, most material can be made to sound significantly louder without the subjective nature of the music being changed in any apparent way, and when you're mastering for CD production, that's what's usually needed. How many times have you made a CD of your own mixes only to find it sounds rather quiet next to a commercial album? That's because the professionals use elaborate limiters like the L2 Ultramaximizer.

Given its professional spec and performance, it's no surprise that the L2 Ultramaximizer is not a 'cheap and cheerful' processor — it's a very serious and capable mastering tool with a price to match, but there's no arguing against the fact that it delivers all that it promises. I imagine that the majority of project studio users working entirely in the computer domain could get by quite happily with an L1 plug‑in. However, if you still rely on hardware, then a more flexible and high‑quality hardware unit for both analogue and digital mastering than the L2 Ultramaximizer will be hard to find.

Noise‑shaped Dithering

Any processing applied to a digital signal tends to require an increase in digital word length (number of bits per sample) if sound quality is not to be compromised. However, in order to output a signal at its original word length, some of these extra bits will need to be removed. For example, if you ended up with a 19‑bit signal and you needed a 16‑bit output, you'd have to get rid of three bits. Simply truncating (throwing away) the least significant bits works, but it also throws away some of the dynamic range, the practical result of which is that very low‑level sounds become progressively more distorted.

The process of 'dithering' is used to get around this problem by adding a small amount of white noise to the signal before the word length is reduced, which has the effect of allowing the signal to fade down into the noise floor much more smoothly than if the data was simply truncated. A further improvement can be achieved by adding this dither noise only to parts of the audio spectrum where the ear is relatively insensitive — mainly up above 15kHz — thereby improving the subjective overall signal‑to‑noise performance. The IDR system works in much this way, preserving the integrity of low‑level detail within critical spectral regions and leading to an improved stereo image and more smoothly fading reverb tails.


  • Maximum loudness with a minimum of side effects.
  • Extremely easy‑to‑use control layout optimised for mastering.
  • IDR noise‑shaped dithering options.


  • Relatively high price.


Extremely high‑quality mastering limiter which is flexible, ergonomic and very easy to use. However, tools as professional as the L2 Ultramaximizer only come at a price...