You are here

Steinberg Wavelab 1.5

Audio Editing Software & DSP Plug-ins By Paul White
Published February 1997

How do Steinberg's DSP software plug‑ins for their WaveLab software, now at version 1.5, compare with the industry heavyweights? Paul White unearths his nastiest recordings to find out...

Steinberg's general‑purpose PC‑based digital audio editing package, WaveLab, has now reached version 1.5 (see the full review of version 1.01 in SOS August '96), and is capable of editing samples or audio files of up to 24‑bit resolution, dialogue, music and sound effects (though there's no dedicated sampler support). The use of a floating point algorithm produces a huge internal dynamic range, and the program is also very fast, very stable and comes with a number of useful software plug‑ins at no extra cost (details of which in a moment). Version 1.5 also allows plug‑ins to be used in real time, but unless you have a fast PC, the more DSP‑intensive stereo processing will probably have to be done off‑line. However, it is possible to run several of the more routine plug‑ins simultaneously, and when processing off‑line, you can apply several processes simultaneously, or even batch‑process files, courtesy of some very clever background processing.

Further plug‑ins are available as options (ie. you need to pay to obtain these; they don't come with WaveLab), and some of these are very serious tools indeed — hence this review. Optional plug‑ins so far include the DeNoiser, the DeClicker, the Spectralizer harmonic enhancer, the Loudness Maximizer, and the brand new Magneto, which is designed to emulate tape saturation. One of the plug‑ins supplied with WaveLab is the Grungelizer, which is a real bonus. As the name suggests, this module 'grunges up' your audio, and is perfect for recreating vintage vinyl effects, complete with crackles, hiss and hum. I don't want to appear cynical (no, honestly, I don't!), but it occurs to me that Steinberg might have collected up all the sonic rubbish that their other plug‑ins take out, and then repackaged it so that we can put it back in again! Actually, I'm kidding — this plug‑in creates stunningly authentic vinyl sounds, and according to Steinberg, the dance market has been crying out for something like this for a long time. If this is an effect you need, this plug‑in is the way to achieve it!

Wavelab v1.5

The main difference between WaveLab v1.5 and its predecessors is the RealTime engine, comprising a Master section (see Figure 1) with six series‑connected, virtual insert slots. The window also has nice meters, sophisticated dithering (see the 'Dithering About' box), clipping counters, a real‑time level display, and an indicator for detecting dropouts of as short as a single sample. The manual states that users with a Pentium PC slower than 90MHz can turn off the Master section if necessary, but depending on the plug‑ins used, I'm told you can even get by with using a 486. If you try to run the Master section with a 486, you get a warning telling you that you're underpowered, but you'll be allowed to try anyway, and depending on the plug‑in you want to run, you may well get away with it.

Any internal or external plug‑ins can be dropped into WaveLab's six effect slots, and the number that may be used at once depends on the power of the PC running the system, and on whether mono or stereo processing is needed. A Pentium 166, or, better still, a Pentium Pro 200 is recommended for serious real‑time use involving the more demanding plug‑ins, but in the context of a professional environment, such an expense is not unreasonable. However, you can do a real‑time preview in mono, and then process the file — which doesn't take nearly as long to do as you might think — and what's more, you can preview several plug‑ins running at once, before processing them in one operation! This is in direct contrast to Digidesign's TDM plug‑in system, where if you run out of DSP farm power, you have to process the file using some plug‑ins, shut those down, then reprocess using whatever other plug‑ins you need. I'm also told that running under Windows NT instead of Windows 95 makes the processing some 35% more efficient!

Bundled with WaveLab on the master CD‑ROM are nine plug‑in effects: AutoPanner, Chorus, Echo, EQ 1, ReSampler, Reverb, Leveller, Tools 1 and the Grungelizer. If the purpose of all of these is not self‑evident from the titles, don't worry, as I'll cover their functions in just a moment. The optional plug‑ins are protected by a hardware dongle, though Steinberg are thinking about changing to a disk‑based system. As it is, if you want to run all the optional plug‑ins, you either need a short ribbon cable to allow the dongles to hang down the back of your PC, or you need to knock a hole in the wall to allow them to poke through into the next room!

Once the plug‑ins are dropped into their slots — actually, it's more a case of calling them up from a pull‑down menu than physically dropping anything — they operate as if they were connected in series with the input, feeding module one and the output coming from module six. However, whether or not you can use all six slots in real‑time or not depends on what plug‑ins you want to use — for example, you can have several EQs and a sample rate convertor running at once with even a fairly basic machine. The DeClicker, Loudness Maximizer and Magneto, on the other hand, really need a Pentium 166, though I'm informed that the Loudness Maximizer and DeClicker manage OK on a Pentium 133. The two plug‑ins that really stretch your system are the DeNoiser and the Spectralizer, but even if you go the whole hog and buy a Pentium 200 to run them in stereo and in real time, the total cost is still way below that for any comparable 'audio cleanup' system.

Other than the real‑time window and the bundled plug‑ins, there are few obvious mods to the basic program itself, other than the addition of a Windows menu bar for accessing the RealTime Engine functions and the ability to read AU (Sun and NeXT) formats, as well as RAW audio files. From my own point of view, WaveLab desperately needs a playlist feature, and if anything, the new plug‑ins just make this more important. However, at this point, nobody's really sure what direction Steinberg plan to take with this product in the future.

Bundled Plug‑Ins In Detail

    The AutoPanner is a simple device designed to pan audio from left to right and vice versa under the control of a low‑frequency oscillator. The modulating waveform may be a sine wave or a pulse wave, and in addition to LFO frequency, you can also set the width of the pan, and control the two output levels independently. There are no fancy sweep sync options, and no MIDI control.
    This is the familiar studio effect, but with the addition of a Glimmer parameter that adds stereo panning to part of the treated signal. The user has control over depth, rate, feedback and delay, as well as independent control over the level of the feedback signal. Stereo spread, mix and output level may also be adjusted. The result is a very usable and versatile chorus, though I can't envisage using Chorus at the editing stage very often.
  • ECHO
    Another familiar effect, this time with control over delay, feedback, balance and volume. Each channel may be set up separately to create a true stereo effect, and it's possible to link the channels so that the output of effect block one feeds the input of effect block two.
  • EQ 1
    Here you get variable‑frequency high and low shelving filters, plus a fully parametric mid section, and in case you need more control, you can load the plug‑in into two or more slots, provided your PC's processor can handle the strain. This way, you can set each one to work on a different part of the audio spectrum. Though digital EQ can lack character, this one is very positive, and does much as you'd want it to.
    Think of this module as a digital volume control, able to increase as well as decrease the level of an audio signal. The range is from ‑48dB to +12dB, and the purpose of the module is to make up gain lost by other processing, such as EQ.
    On the face of it, this is just a sample‑rate converter, but it has the advantage of being usable in real‑time as well as off‑line. The sample rate is adjustable from 11.025kHz to 48kHz, and depending on the application, you can choose from three quality modes to save processing time at the expense of quality. Mode is used to select the target sample rate if it is to coincide with one of the standard rates supported by your soundcard.
    Lexicon aren't going to lose too much sleep over this one, but it's good enough to be useful nevertheless. Nine parameters are available for user adjustment, including room size, decay time and several early reflections‑related functions. At some settings, this plug‑in can sound a bit 'ringy', but it's useful for adding ambience to dialogue or music. Considering no additional DSP cards are needed to run it, it's amazing that it's as good as it is!
  • TOOLS 1
    This is an odd but useful inclusion, capable of independently inverting the phase of either channel, or working in one of two M&S (Mid & Side) modes. One mode deconstructs stereo signals into their middle and side components, exactly as you'd get from a middle and side mic, while the second option takes the output from a middle and side mic array and converts it to conventional stereo. Interesting experiments include splitting a signal into its M&S components, EQ'ing or adding reverb to the side components only, then using the M&S decode facility to put the signal back into stereo.
    This plug‑in is designed to make pristine recordings sound like vinyl by adding crackles (with a choice of 33, 45 or 78rpm settings!), hiss, distortion, vintage EQ coloration and mains hum at 50 or 60Hz. I wasn't entirely convinced by the harmonic content of the mains hum — it wasn't quite angry enough — but everything else is absolutely authentic. There's even a time control which can set the playback character to anything between 1900 and the present day! You must try this — absolutely any audio can be converted to sound like a badly‑looked‑after vinyl record played back with a chipped stylus! Uniquely, this plug‑in has a brown front panel, presumably out of respect for its effect on the sound. Aside from its obvious applications in dance music and sample creation, the effect is also useful for post‑production 'futz', where it may be necessary to fake a vinyl sound for film or TV. This would be a great plug‑in, even at a premium price, but to get it bundled for no extra cost is a nice surprise.

Optional Plug‑Ins

    All of the optional plug‑ins can be used to process files of up to 24‑bit resolution. The Loudness Maximizer combines level change with automatic, intelligent dynamics control to make a signal sound as loud as possible (and to make use of all available headroom), but without introducing undesirable side effects, such as gain pumping or dulling of transients. As with most of these plug‑ins, the controls are deceptively simple. One fader sets the desired signal gain, and alongside is a meter showing the possible gain increase that can be achieved without clipping. Controlling the dynamic side of the process is the Density slider, and next to that is a meter showing the 'Desired Gain Done'. In effect, Density controls the relationship between the compression and limiting aspects of the process, and at higher settings, more of the gain reduction is achieved by compression than limiting, resulting in a higher average signal level. As its rather clumsy name implies, the Desired Gain Done meter shows to what extent the process has been able to provide the desired increase in gain.

A Boost button adds a further 2dB of increased loudness, regardless of the other settings, and the Limiter section is switchable in 10 steps between soft and hard modes. Hard settings work well with aggressive, rhythmic pop music, while 'softer' values are better suited to classical and acoustic music. High‑resolution output level meters are included, showing the output level to an accuracy of 0.1dB. User settings can also be stored and recalled for future use. As well as making mixes sound louder, the process is useful for treating 16‑bit multimedia audio files before reduction to 8‑bit.

This plug‑in certainly makes signals seem a lot louder and punchier without increasing the peak level at all, and what's more, you can set the target peak level at any value you like. Most impressive is the lack of traditional compression or limiting artifacts — something to do with the cunning 'look‑ahead' algorithm. To do this with analogue, you'd need a chip that could see into the future! I suppose the nearest approximation to this plug‑in is the L1 Level Maximizer from Waves, and I have to say that I feel the L1's controls are a little more logical — but that could be because I've been using it for quite a while now.

    Spectralizer is really a harmonic enhancer that works entirely in the digital domain, providing independent slider control over added second and third harmonic content. As with a conventional enhancer, a frequency control allows the user to set the frequency above which harmonics are generated. As the second harmonic is one octave above the fundamental, setting a filter frequency of 3kHz, for example, means that the lowest added harmonic will be 6kHz.

A Kick switch adds extra processing to transient sounds, and if you're not exactly sure what's going on, a neat block diagram of the process can be called up next to the control window. All the variables in this module are controlled by sliders rather than by rotary controls, and once again, favourite settings may be stored.

Subjectively, the process is similar to an analogue enhancer, but it's quieter and more controllable. The sound remains warm and detailed, but there's no separate low frequency control, which might have been useful. This is a useful tool to have available during post‑production audio sweetening, where you don't want to leave the digital domain. It's also useful for brightening up samples.

    Noise removal software has been around for some time, but the results that can be achieved vary wildly depending on the type of process being used. No system is completely successful at tackling all types of noise at once, so the process tends to be split up into de‑clicking, de‑noising, and de‑crackling. De‑noising is used to remove broad‑band noise such as tape hiss, while de‑clicking tackles individual spikes, such as digital glitches or vinyl scratches. De‑crackling is probably the most complex and specialised process of all, and is intended to clean up dense surface noise from vinyl. De‑crackling is not yet an option for WaveLab, though dramatic improvements to vinyl recording can be made using the de‑clicking and de‑noising processes. Perhaps now I'll get around to putting my old vinyl records onto CD‑R before they're completely unfit to play...

No specific details are given as to how de‑clicking works, but the process really comprises two parts. The first task is to decide what is a click and what is a meaningful musical transient, and I'm told that a certain amount of modelling is used to identify genuine clicks. A user‑adjustable threshold provides the user with control over the severity of the process, while Audition mode allows the scratches to be heard in isolation. This is immensely useful, as it's quite obvious whenever meaningful material is wrongly identified. Once you're sure that only scratches are being picked up, you can run the system in real time or apply the process to the file in question, and as if by magic, the clicks are removed, and their space filled with plausible material calculated from whatever is happening either side of the click.

To further fine‑tune the process, the user can choose from four levels of processing, which give priority either to the amount of noise removed or to the quality of the finished audio, and there are three further settings for Old, Standard and modern material, roughly equating to 78s, vinyl at 33 or 45 rpm, and material with digital glitches. Of course, you can try any of these settings on your material to see what works best.

How well does it work? Better than you could possibly expect! Quite badly‑scratched vinyl cleans up very nicely, digital spikes or clicks are dealt with, and even clipping distortion is made more tolerable. Setting up is extremely easy, thanks to the Audition mode, and the quality of restoration is quite simply amazing. If there are better systems around, they can't be that much better, and yet most are a lot more expensive! As with the other plug‑ins, you can run in real time or off‑line, and settings may be saved.

    De‑noising is invaluable for reducing the level of tape or equipment hiss, especially if you're remastering old recordings, but most simple systems suffer from noticeable side‑effects. For years, I've been using Digidesign's DINR, but realistically, you can only expect a maximum of 4 or 5dB of noise reduction before you start to hear the noise floor being modulated by the noise removal filters.

Unlike DINR, which really needs a sample of 'noise only' recording before it can start work, DeNoiser uses a more advanced system based on an intelligent, adaptive algorithm to constantly reappraise the noise. When the file in question is run, a small window shows the dynamically varying noise spectrum, over which is superimposed a threshold line. This is the main user adjustment, and once the threshold is pushed just above the central portion of the noise spectrum level, processing begins. There are only three user controls — one for the threshold level, one for the amount of gain reduction and one for ambience. The Ambience control is designed to stop the process robbing ambient sounds of their character during the decay of the sound, and I have to say, it really works.

Once again, files of up to 24‑bit resolution may be processed, and two different settings can be stored for direct comparison using the A and B buttons. Used carefully, the quality of processing is excellent, and in practice, you can get between 12 and 20dB of hiss reduction without messing up the sound, though the exact figure depends on the quality of the material being processed. Interestingly, when I tried the process on a sample of vinyl record recorded onto DAT, DeNoiser wiped away most of the steady‑state surface noise, leaving only the scratches and the worst of the crackles. As you'd expect, the DeNoiser isn't designed to salvage horrendously noisy material, but even salvage jobs work out better than they have any right to. On the other hand, decently‑recorded material with only moderate noise contamination cleans up beautifully.

    This plug‑in turned up in late beta form, just before the review, and with no manual. It was also the only one of the plug‑ins that wouldn't run without glitching in stereo/real‑time mode on a Pentium 166, but on contacting Steinberg, I was told that this was due to the beta version not being fully optimised, and that the production version would be significantly faster.

The controls are pretty much self‑explanatory. Featuring two nicely vintage VU meters, complete with extra‑large red overload areas, Magneto is designed to emulate the effects of tape saturation at both 15 and 30ips. The meters can monitor the input level, the output level, or the virtual to‑tape level, and an HF control allows a little fine‑tuning of the top end of the spectrum. Other than that, it's simply a matter of overdriving the virtual tape by whatever amount produces the right subjective result, and though the process is fairly subtle, it's actually quite authentic.


WaveLab has gained a lot more credibility thanks to the high‑powered optional plug‑ins, developed for Steinberg by Spectral Designs. All the bundled plug‑ins are clean and competent, (other than the Grungelizer, which is delightfully dirty and competent!), but for my money, the DeClicker and DeNoiser take the top prizes. These tools can produce genuinely professional restoration results, yet they remain simple to use and, above all, affordable. As I said earlier, I've been using Digidesign's DINR for several years now, but DeNoiser makes it look like a packed lunch! And, while I personally don't often have scratchy vinyl to deal with, the DeClicker is exactly what you need for sorting out those DAT transfers that have accumulated inexplicable clicks along the way.

The Grungelizer is worth a mention for its audacity, the Leveller is just what's needed for squeezing the last drop of subjective level out of an audio file without making it sound compressed to death, and the Magneto can add analogue warmth to a DDD recording. In all, this is a beautifully‑presented, very slick suite of programs, and though the processor‑intensive DeNoiser can only be run in real‑time if you have a 166MHz Pentium computer or above, I can't blame the software — processing of this quality takes power, and if you have to wait a while for off‑line processing, that's just the way it is.

If you need powerful, affordable audio processing, then you'll be doing yourself no favours by ignoring WaveLab and its plug‑ins.

Using The Plug‑Ins

The Master section and the plug‑ins (other than the Grungelizer) all appear on‑screen as 'virtual' red anodised control panels, and a chunk of audio data — known as a Wave — may be processed in real time or by 'Applying' the process or effect(s) from the Master section. Real time means exactly what it says — you can feed a DAT tape into the input of your soundcard and a treated version pours out of the other end. However, be warned that the default mode of real‑time operation is mono, so as to allow glitch‑free previewing of signals on less powerful PCs. To select stereo preview mode, you have to remember to hold down the Control key while loading the plug‑in. While accepting the reasoning for this, I do feel a large mono/stereo warning sign somewhere on the screen would be in order. When a file is processed, however, it is always handled in stereo mode, regardless of whether it has been previewed in mono or stereo.

In Apply mode, you set up the parameters for as many effects or processes as you want to use, and then, when you apply these, a new file is created containing the results of this processing and the master fader settings, plus any dither option selected. There's also the facility to batch‑process Waves if you're applying the same process to a number of files. A Monitor window shows, amongst other things, how much of your CPU power is being taken up by the current process, and when Live Input is selected, a nice little picture of a mic appears.

Each of the six processor slots in the Master section has its own Bypass button, Mono button, Solo button and FX button, the latter being a quick way to access the control panel of the module currently occupying that slot. The individual plug‑ins also have Bypass buttons, and most have clip warning indicators, but be warned that unless a module is bypassed on the Master section (or removed from its slot), it still uses up as much processor power as if it were actually working.

Control over plug‑in parameters is via on‑screen knobs or sliders (controlled by dragging the mouse up or down), and via click switches. The virtual pots on the plug‑ins are adjusted by sliding the mouse up or down while holding down the mouse button, but I found this a little jerky on occasions, presumably because of the amount of processing going on at the same time.

Dithering About

The dithering option in WaveLab's Master Section is there to increase the dynamic range of signals that are being truncated to fewer bits or are being processed in some way. It isn't a new principle, and works by adding mathematically‑generated noise to the signal to 'dither' the data so that low‑level signals can be heard disappearing politely into and below the noise floor rather than simply expiring in a crunchy mess. The amount of noise added is very small, and two basic types are available: Type I is general‑purpose, while Type II emphasises higher frequencies than Type I. There are also three noise‑shaping options, designed to distribute the added dither noise into parts of the spectrum where the ear is least sensitive.


  • Fast and solid.
  • Excellent plug‑ins (both bundled and optional), especially the DeNoiser and DeClicker.
  • Sensibly priced.


  • Real‑time processing requires a fast Pentium machine.
  • WaveLab itself doesn't yet have all the necessary tools for album compilation work, and also lacks sampler support.


An impressive editing package that brings true high‑end processing into the mid‑price bracket. It's also very easy to use, and most operations are surprisingly fast.