Daniel Weiss' hardware helped convert mastering engineers to the joys of digital processing. Is it still equally desirable in plug-in format?
On any list of classic digital mastering equipment, the name Weiss has to be very near the top. When the Weiss Gambit series of mastering processors first went into production in 1998, the EQ equaliser and DS compressor were regarded by many mastering engineers as the pinnacle of digital processing. Weiss developed them further over the next decade, and they are still made today. My mastering facility at Philosopher's Barn bought a Weiss EQ‑MK2 15 years ago, upgraded it to the full Linear Phase/Dynamic EQ spec a little later, and it is still in regular use.
The sister unit, a de-esser and broadband compressor, was also on our most wanted list, but the price tag was a barrier when we were starting out, and it remained only an aspirational purchase decision. The current purchase price of the Weiss DS1‑MK3 is still around £8500 — so the announcement of a plug-in version for around 5 percent of that price has understandably aroused attention. Adding to the anticipation is the fact that Softube's DS1 plug-in uses the original code and is endorsed by Daniel Weiss himself, who says that it sounds no different from the original.
The Weiss hardware processors are an ergonomic dream, with nicely weighted, touch-sensitive, switched rotary controls for the main parameters, and square push-buttons for menu items that light when active. They are very easy to use, and with two banks of 127 numbered saved parameter states, they are easy to recall. It seems to me that the user experience was a large part of the appeal of the Weiss units, especially for mastering engineers whose processing chains were predominantly analogue — there are very few other hardware digital devices that come even close. And ergonomics is not just about comfort, it is also about performance. For example, from a purely sonic point of view I might just be able to replace the analogue mastering chain I have used for over a decade entirely with plug-ins, but both physiologically and psychologically, the difference in operation would very likely have an effect on the result. The reason for mentioning this will become much more clear later in the review when I look at the additional plug-ins Softube have derived from the Weiss algorithms, but the point is simply that identical digital code does not yet provide for an identical human response. And, of course, music engineering is a very human activity.
For the central DS1 plug-in Softube decided not only to port the original code line-by-line, but also to adopt a GUI which hardly deviates from the original in appearance, and which, sadly, is equally resizable. Clearly this is intended to recall the experience of using such an iconic piece of kit, but it's risky to ignore the differences between hand and mouse, rack space and screen space, so it's good that Softube have attempted to address this with additional direct-drag modes of control.
The Softube plug-in actually features slightly more controls than the hardware. At top right are the original six rotary controls for the non-standard compressor settings, which govern a crucial aspect of the performance of the DS1. In addition to the standard attack setting, these include Attack Delay and two different release stages called Fast and Slow; the side-chain processor measures both peak and RMS levels and then 'decides' on the basis of their ratio which release speed to apply. The Average control sets the time window over which the RMS is measured, so faster Average settings mean that the processing leans more to Fast release. There is no substitute for trial and error for finding the optimum setting in each case, but after a while, most operators find a favourite, predictable, narrower range. The dual-stage release is not in itself unique — both the digital TDR Kotelnikov and the analogue Pendulum OCL2 (see below) feature similar architectures — but this amount of fine manual control between them is much less common.
Six, more widely spaced, rotary controls on the lower right add standard knee, threshold and ratio controls, along with Shape and Centre Frequency Bandwidth settings for the band-selective compression. This, in effect, is an extremely useful generalised form of de-essing: you isolate a certain range of the frequency spectrum with high- and low-pass filters and a variable band-pass filter, and then apply compression to it. It's a powerful tool both for attenuating problem frequencies and tightening and augmenting others. Because the make-up gain control on the far right affects only the gain of the selected frequency band, this makes possible more intrusive and 'creative' processing than might actually be desirable at the mastering stage, but for more restrained use, this is obviously a good thing. The Softube DS1 is no longer unique in this ability — which is why the reuse of the marketing blurb for the original, which compared the flexibility of the Weiss to relatively lumbering analogue units, seems a tad irrelevant — but it is very good at it.
Supplementing the rotary controls is a staggered column of five jelly-tot 'push-button' squares in the middle left, which toggle the Options menu, Waveform view, and three targets for the Gain control: Input, Output and Limiter. There are three limiter options: the original, as found in the hardware, and two added by Softube. I didn't use them much in the review sessions: they do not have any parameter control and in this respect the separate Softube MM‑1 limiter plug-in, based on the Weiss algorithms but with much more user control, fares much better.
Along the bottom left there is a large green bypass button, alongside a row of 10 more jelly-tots which toggle their named functions on and off. Most of these controls will be familiar from other modern software compressors. The original Weiss has fewer buttons and slightly different distribution of the functions, but as the changes in the software mean fewer functions are hidden in menus, they are all to the benefit of the software UI.
The bulk of of the left‑hand side is devoted to a screen which, on the hardware, displays the compression settings and gain reduction. In the software, however, this can also be switched to a new Waveform view which provides a nicely clear graphical presentation of the compressor action. Oddly enough, a default setting in the software means that once a 'touch‑activated' rotary control has been released by the mouse, its numerical setting will disappear and only then be visible in a portion of the left-hand screen; however, a 'Bob Ludwig' mode, in which the current parameter values are permanently on display, is a Setup option. This seems to be an obvious choice to invoke, especially if the user is intending to work with the Waveform view more or less permanently in place. During the review, however, I kept the Waveform view off because doing so enables another nice new software feature: the direct manipulation by mouse drag of the various parameters, which to my mind is a much better way to interact with digital processors than 'rotary' controls.
I have borrowed and rented Weiss hardware units to use on a number of mastering sessions in the past, and I decided to revisit some of those sessions from the archives and, using the same settings, see how the new plug-in compared with what I had already done. Nearly all of those projects were mixed and mastered at high sample rates, with the hardware DS1 used mainly for de-essing and other band-selective compression, most of it relatively wide-bandwidth, high-frequency work. I had only rarely used it as a broadband compressor: where I had gone to the trouble of obtaining a unit it was because I was interested in the selective frequency possibilities, and in particular the isolation of the Middle and Sides signals, a function which only arrived with the final (MK3) version of the hardware processor. In none of them did I use the parallel function.
The watchword for the DS1 seems to be smoothness, and that is how I had used it in these projects: a lot of gentle dynamic control with small ratios, high thresholds and relatively slow time constants. I wasn't expecting any difference in sound between the hardware and software results, and there wasn't any. But what was interesting to me when revisiting these projects was how often found that I would have done things slightly differently. In particular, in a number of cases I would have probably reduced the overall amount of compression. As a lot of these projects generally involved exposed and sparse textures, this was a surprise. But it shows two things: the first is the extent to which even when the DS1 is slightly overcooking things, it still tastes extremely nice; the second is the influence of a new breed of flexible, full-featured and transparent compression plug-ins which have arrived on the market over the past few years. With their high standards in mind, when using the DS1 in its new incarnation I found it possible to moderate the processing a touch and still be satisfied with the results, while at the same time finding a greater transparency.
One earlier project I revisited was an electric folk album featuring a female singer who had multitracked her harmony vocals. Listening back, I felt these were slightly over-compressed, but when I then tried to correct for this as before with the new software, I found it much more difficult to do. In my notes I had a comment from the singer on an earlier master of the same track in which she complained that it had made her voice sound too much like the Beverly Sisters.
I then decided to try a different kind of comparison, by going back to much more recent sessions where I'd used a compressor from my current toolkit. The original Weiss marketing had made much of the challenge from analogue compressors, so I chose one example of that, and also a modern digital compressor. These comparisons are not meant at all as shootouts but rather to illuminate the strengths and weaknesses of the Weiss. The Weiss should really shine in situations where you'd use a digital compressor, so I gave it an outing with two quite different tasks: a de-essing task for a solo vocal track, and then broader split-band duties for overheads in a full mix.
Setting up the de-ess and band-selective processing is a breeze: using the monitor function to audition the band to be processed, for instance, it is pretty easy to dial in a broad processing band and then refine it by adjusting the centre frequency and the bandwidth control. In theory, the Softube DS1 is not as versatile as some more modern designs, but in practice, I didn't miss any of the extra flexibility those designs offer.
The first example was a jazz project mastered from stems, where the vocal stem suffered from the devil's brew of proximity effect and sibilance. In the original mastering I'd used the Sonoris Mastering EQ plug-in to deal with the former and DMG's Essence for the latter. For this review, I retained the same EQ, but replaced Essence with the Softube DS1, at first set up to replicate the DMG settings as closely as possible, and then fine-tuning it for the best subjective results.
We'd been very happy with the results from Essence, and I was now really happy with what the DS1 delivered, but it was interesting that the results were nonetheless audibly quite different. Essence had extracted everything that was problematic, but otherwise left the original apparently untouched; DS1 had also dealt with the problem, but although it hadn't adversely affected the original, it had nonetheless changed it slightly. There are a lot of digital de-essers which introduce far less innocent differences, and both the DS1 and Essence are streets ahead of those, but whether the change from Essence to Softube DS1 is an improvement or otherwise is likely a matter of taste. I suspect that if you only use the Softube DS1 you'll find it to provide superlative de-essing, but if you already use something in a similar class to Essence you might find yourself having to make a choice between accuracy and another kind of 'rightness'. In fact, if you have the wherewithal, you could buy both and send yourself nuts with daily indecision as you try to decide just which one to use this time.
The second example came from a favourite mastering project, the second Szun Waves album. This is an extraordinary blend of improvised electronica with jazz drums and reeds, which sometimes presented mastering challenges in terms of the textural complexity and dynamic range of all the instruments. One track had an extended explosive final section in which the drum overheads had become a bit overbearing and splashy in the mix and which needed to be controlled without compromising their role in keeping and elaborating the basic pulse. In the original session, I'd used a Pendulum OCL2 tube compressor with a high-pass filter at about 300Hz on the side-chain and a lot of careful parameter choreography. The Pendulum has what Brad Blackwood once called "that sound", and as I've seen recent forum comments that the DS1 plug-in also has "that sound" (though it might be a different 'that') I was intrigued to see what it would make of the same material.
Doing so brought up an important point: using the DS1 for the first few times demands caution. Because it does do something nice to the sound almost immediately, and because it seems that a decent enough setting can be dialled in pretty quickly, there is the danger of overlooking the down side or, conversely, of not realising that the good can be made even better with further attention. This is basically what happened here: hearing the overheads as a 'problem' before, I was initially greatly impressed with how the DS1 solved that problem, but when listening through to the dynamics of the whole track I realised that the 'build' of the explosive ending wasn't as musically engaging. So, I started again, trying to get the DS1 to maintain the essential control and then fine-tuning the strident/velvet dimension. The result was acceptably close, but only after a lot more fine-tuning of the frequency window and the average setting.
The practical take-home from both of these examples is that the DS1 plug-in is a very nice piece of kit, but I would recommend that potential purchasers use the demo period to the max. Extremes of reaction from most users to the plug-in will probably average out over that period, with the initially gobsmacked perhaps more aware of the limitations of the processing (such as they are) and the initially sceptical more aware of the subtleties of processing available as they get more familiar with its ways of working.
The first version of this review was already completed, and I'd sent the copy to my editor and closed my computer, when he asked if I could add a few words about the interesting new package deal which Softube were now offering for purchasers of the DS1 software. Not content with offering the classic Weiss algorithms in the classic Weiss interface, the developers have also created separate Compressor, De-Esser and Limiter plug-ins. At first I thought this was a bit like releasing too many singles from the same album, but as I explored the individual items further, I found some genuine surprises.
Some of these arise from differences between the new plug-ins and the DS1, such as the dual EQ points in the new De-Esser, while others emerge from the different presentation of the parameters and their control. For example, when I dug a little more deeply into using the new Compressor I found that I was able to more quickly get good (and perhaps even better) results than with the original interface. The fact that ergonomics can bring about a change in the sonic results does not surprise me, but the point here is that even after I had checked with Softube to confirm that this Compressor used exactly the same Weiss algorithms, I found myself hearing changes slightly differently and so getting better results — not when I tried to replicate settings from other projects, but when I started with new material from scratch.
Buying all three plug-ins separately actually costs slightly more than buying the full Weiss DS1-MK3 package, which includes the three separate plug-ins for free; I don't really understand the marketing strategy behind this, but it seems a pretty good deal! However, it wouldn't be so tempting if the three separate processors just replicated the respective bits of the larger plug-in. Fortunately, they differ in important enough ways to make it a real deal. In essence, the new De-Esser adds functionality, the Compressor simplifies existing functionality and adds a vastly superior software UI, and the MM‑1 limiter does a little bit of both.
At first glance, the De-Esser seems to be a greatly stripped‑down version of what is in the DS1: no full-band operation, no low-pass, no dithering or auto-gain, no Safe Limiter, a fixed Preview setting, no RMS detection, and a narrower operational range that does not go below 2kHz. But weighed against all of those changes are the doubling of the operational bands to two, and the new modes of parameter control. The twin‑band approach is an exceedingly useful development: very often the harshness in a vocal is not conveniently located at a specific frequency, but at related frequency points (or ranges), and this alows both to be dealt with in the same plug-in on the same screen. In a single process you can nicely tame, say, 3.5 and 7 kHz: related frequencies, but likely with very different de-essing needs. And the further bonus is that the plug-in screen is fully interactive, allowing you to change the frequency, threshold, range and width of the filters by clicking and dragging on the dedicated area of the screen.
It is this aspect of the user interface that transforms the experience of using the Compressor/Limiter as well. Again there are similar simplifications of the controls and options — no low-pass filter, a simplified form of parallel processing and an auto-release in place of the three controls in the original DS1 — but the ability to control important parameters by clicking and dragging struck me as a significant usability enhancement. Of course, other plug-in manufacturers such as FabFilter and DMG also incorporate this type of interaction, but in the context of using what was hardware in a new software situation, it's a clear improvement over the DS1‑MK3 plug-in.
As to down sides, there are three main negatives that I encountered. First is cost: if the main criterion is value for quality, not value as compared to the cost of the hardware version, then this is an expensive plug-in. Compared to modern competitors of equivalent quality, it becomes even more so. Of course, this is not the only way to qualify the cost, and the enticements of the new package might take a lot of the sting out of it.
Second is the unfortunate preset management (see box), and third is the documentation. Softube's user manual is mostly a line-by-line copy of the original from a decade ago. Although the few software/hardware differences are duly noted, and one typo from the 'Factory Presets' section corrected, no attempt has been made to provide an update to better meet the possibly very different skill sets, expectations and needs of a more modern user base. The most inventive part of the new manual is a splash picture of a mock-up of a non-existent 'hardware unit' which sports all of the controls of the new software.
This strikes me as a missed opportunity, especially as Softube clearly expect the new user base for the software to expand from professional mastering engineers to encompass professional mixing engineers and probably beyond. There are, for example, no suggestions for the use of the band-selective mode, and even on parallel compression — a feature implemented quite early in the evolution of the Weiss units — the manual still only has a single 15‑year‑old quote from Bob Katz, followed by a brief description of his particular way of approaching it. This lack of concern for the user is not evident in the original Weiss manual, which contains a three-page addendum containing 14 clear graphics which, to quote Daniel Weiss' words, "convey a better understanding of the dynamic parameters of the Gambit DS1 De‑Esser/Compressor". And indeed they do, so it's lucky that they can still be found in the PDF of the hardware manual on the Weiss web page (www.weiss.ch/files/downloads/ds1/DS1-MK3-Manual.pdf).
The Softube site prominently features a quote from Bob Katz saying that the original unit was "the most transparent, refined, flexible and least 'digital-sounding' dynamics processor I have ever used". But that review dates from 15 years ago, and 'digital-sounding' is not now what it was then, if indeed it is now anything particular at all. A processor which, 15 years ago, was able to sound the least digital might now be over-compensating a tad. It was clear from my experience with both hardware and software that the DS1 does in fact have something of a sound or a character of its own. Especially in broadband mode, the DS1 imparts a kind of velveteen texture to the results that other users have called 'polished' or 'professional' or 'expensive'. It is not quite an effect (at least, not at higher sample rates) but it means it's unlikely to work equally well on everything. For quite some number of purposes, though, the distinctive sound that the DS1 offers is exactly what is required, and in most circumstances otherwise very difficult to achieve. It remains a complex and hugely powerful processor, and the additional plug-ins that Softube have created using the same algorithms are a great help in harnessing that power.
There are no other plug-ins using the Weiss algorithms that I know of, but if you're looking for a fully featured, mastering-oriented dynamics processor that will tackle de-essing, consider Sonoris Mastering Compressor, DMG Essence and the very affordable TDR Kotelnikov GE.
Softube's manual for the Weiss DS1‑MK3 plug-in says that preset management can be carried out either in the host DAW or using a dedicated Preset Collection tool, which is where the ubiquitous Bob Katz presets reside. I find my presets useful as very basic starting blocks to be modified according to musical needs but, more importantly, I also need to save per-song settings in mastering — up to a dozen or so for each processor per project. So having a simple one-click Save Settings button is ergonomically important, a fact which is recognised in all of the software processors I currently use. In Sonoris and DMG plug-ins, for example, presets are easily saved as tiny files with a unique suffix.
The Softube Preset Collection is complete overkill for this, incorporating all kinds of user options from preset descriptions to photos (for some reason). In a situation where I create roughly 12 new settings per project and work on perhaps four projects a week, this would quickly become cumbersome, and despite all the fancy features, I could not find a simple way to store settings for each track within individual project folders.
- A software recreation of a mastering classic.
- A superb de-esser, and a broadband compressor with an identifiable sound which is complimentary to a wide range of material.
- User interface improvements in the three additional plug-ins make a substantial difference and add considerable value to the package.
- Some might find the hardware-like user interface for the main DS1 plug-in unwieldy.
- Manual and preset handling could be improved.
A faithful recreation of the original digital hardware, Softube's DS1 is still one of the best de-essers around. Used as a broadband compressor, it applies a distinctive sheen which can be very appealing on the right source.
$599; Compressor/Limiter $299; De-Ess & MM‑1 $199 each.
$599; Compressor/Limiter $299; De-Ess & MM1 $199 each.