Wavelab 's latest tricks include a revolutionary Spectrum Editor, which allows you to identify problem areas in your material visually before applying extremely precise restoration processes.
Since its first release in 1996, Philippe Goutier's Wavelab application has jumped to the next major revision every couple of years. Each new version has managed to add numerous new features and improvements without sacrificing stability or performance. Given that Wavelab 5 already included more features than many musicians could shake a stick at, some may have wondered if there were any more new features that could possibly be added.
Wavelab 6 proves them wrong, but one new addition may not be as welcome as the others. For the first time ever, Wavelab is protected by a USB dongle. The full version is supplied with a Synchrosoft dongle identical to that used by Cubase SX, Nuendo 2 and 3, Hypersonic and Halion 3. Wavelab upgrades don't include this dongle, but if you already own any of the aforementioned products you can use its dongle to authorise Wavelab 6. Those upgrading without a suitable dongle can buy one from Arbiter for £20 (it must be bought at the same time as the upgrade, though).
The most notable new feature is that each waveform window and associated overview can now be viewed in any of three display modes. The standard mode familiar from previous versions is Wave, but this is now joined by Loudness Envelope and Spectrum. You can choose different modes for the main window and overview, and either zoom each separately, or synchronise them so that (for instance) you can have a spectrum display in the overview and wave displays in the main windows beneath, and they will remain locked together when you zoom in and out.
The Spectrum option displays your waveforms in spectrogram format, with time across the horizontal axis, frequency content rising up the vertical axis, and intensity linked to a range of colours or black/white strength. The Spectrum display options include colour, black and white, logarithmic/linear frequency scale, resolution of the FFT spectral analysis (a compromise between frequency and time resolution), and audio range; as this is lowered, only the 'louder' frequency content appears on screen, which is a great way to home in on transients, for instance.
I've always enjoyed spectrogram views because they show things that may not be visible any other way, such as low-level hums and whistles (which appear as horizontal lines) and what enhancers are actually doing to your audio (changes of colour near the top of the window). However, you don't often get the option of performing edits in this mode, so Wavelab 6 's spectrum editing is a very welcome addition in an application at this price.
Once you've selected Spectrum editor mode, an associated dialogue window appears and the cursor changes to allow you to define a rectangular region on the display. With stereo files, an identical region is automatically created in the other channel. You can either treat the selected regions with off-line Surgical processing, or real-time Master Section processing. Both modes use high-quality linear-phase filters capable of 'infinite' steepness in excess of 1000dB/octave, although you have full control over the steepness and the crossfade time where the filtered portion is merged with the original audio signal.
Surgical processing offers a variety of filtering alternatives. Damp is generally used to reduce levels, with a choice of band-pass, low-pass and high-pass filter options to reduce various unwanted parts of a region. Blur Peaks is easy to understand in the graphic sense, and is a powerful tool when you want to remove certain spectral parts of a sound. Dispersion smears sounds so they are less obvious in the mix, without removing them and leaving a hole in the spectrum, while the four Fade options are great for reducing the impact of transients or gradually changing certain spectral aspects of a region.
There are also copious Copy operations that let you transfer a Source to a Destination region, so you can, for instance, replace a bum note with a good one from another part of the song. By working with limited frequency ranges you can also remove or replace sounds like a few seconds of feedback in a rock concert or a sneeze during a live classical concert. The copious and informative PDF manual provides plenty of operational details, although you'll need skill and a lot of practice to perform such miracles invisibly.
Popular stereo editing packages on the Windows platform include Sony's Sound Forge, the latest version 8 of which was reviewed in SOS June 2005. This is a well-featured and highly regarded editor, but has nothing directly comparable to the new Loudness Envelope and Spectral Display or Spectral Editing tools in Wavelab 6. The new version 2 of Adobe's Audition (formerly Cool Edit Pro, as reviewed in SOS July 2002), however, does include a spectral display mode, and is also a widely used and powerful piece of software.
With Master Section processing you can treat the whole file in real time through a chain of up to eight plug-ins. These plug-ins are either fed from a frequency spectrum defined by the region you select, or from the remainder of the frequency spectrum, while the non-effected part of the spectrum can either be mixed in at the output of the Master Section, or discarded altogether for more radical treatments. The latter mode is also useful when fine-tuning the effected part prior to recombining the two.
While Surgical processing is one of the most sophisticated tools there is for audio restoration, Master Section processing offers a huge range of new creative options, such as frequency-selective reverb, flanging and chorus, auto-panning of different parts of the spectrum and narrow-band distortion of drum loops. By the time I tried out spectrally limited delays and echoes my head was reeling with new possibilities. Occasionally the selected and non-selected parts got slightly out of sync after removing complex plug-ins from the chain in real time, but I'm happy to forgive such a small bug in the delay compensation, especially as clicking on the Play button pulled everything back together perfectly.
After the excitement of Spectrum Editing, the new Loudness Envelope display option seems mundane by comparison. It shows loudness over time in various predefined frequency bands, and up to four curves can be chosen from Main (overall), low-pass, band-pass and high-pass. By studying them you can learn a lot about frequency distribution through a song, or the amount of compression used. However, recalculation of the display for a complete track can take a long time, even if you're just resizing or zooming the window, so if you need to do this, switch back to Wave display first.
The Audio Analysis menu options now also include a Loudness Distribution window that plots the most frequent, rather than the average, loudness level in any file. On this display, most modern finished songs appear as a single horizontal lobe, but unaccompanied vocals and classical tracks display significantly wider or more numerous lobes, indicating a far greater variation in dynamics. You can store up to three different results to compare different tracks.
Another related function is the off-line Level Envelope processor that lets you apply a user-defined envelope to any or all of your file. This is far more versatile than the previous Fade processes, especially with its 'smooth envelope' function that turns lines into curves. Yet another enveloping function that I found particularly useful is the Effect Morphing: you apply plug-in effects to whatever part of your file you wish, open this new window, create a user-defined envelope, and Wavelab then mixes in the most recent (unprocessed) Undo file to fade the effect in/out as you wish. I also found this a very useful way to avoid clicks when you want to process only part of a file.
The previous Normalise process has become Level Normaliser, and is joined by Loudness Normaliser, which works with RMS levels to set a loudness and is thus far more appropriate than the conventional peak level adjustment when you're trying to make individual tracks sit in an album context. There's also a new Pan Normaliser which can balance the peak or RMS levels across stereo channels. Finally, the level/pan meters now have K-Metering options, as defined by renowned mastering engineer Bob Katz, which you use in connection with calibrated monitor levels to ensure repeatable high-quality mastering results without compromising dynamics or squashing transients.
Global improvements in Wavelab 6 include enhanced file read/write performance and the removal of all limitations on file size, plus a noticeably slicker graphic interface and icons. There are also many smaller additions and improvements that prove really helpful. Here are some that particularly appealed to me:
Previous Wavelab versions offered the same sample-rate conversion algorithms as Cubase and Nuendo, but Wavelab 6 includes a new SRC Crystal Resampler that's gaining much praise for its conversion accuracy — it's significantly more transparent, and roughly on a par with Voxengo's highly acclaimed rBrain Pro.
Also, tucked away inside both the Time-stretch and Pitch Correction processing windows is an innocuous new tick box that should bring a big smile to all those who do a lot of work with loops. The 'Use DIRAC processor' option switches in possibly the best available time-stretch/pitch-shifting algorithm available on the market today, albeit at the expense of longer processing times. There are six variations on the DIRAC algorithm for dealing with different kinds of source material such as instruments, voices, classical music, percussion and so on, and I found they imposed significantly fewer artifacts on long stretches than the previous algorithms.
Also new in the process menu is a Pitch Quantise option that works on monophonic lines to correct intonation problems; you can define the reference frequency, pitch tolerance and slur time. It's a sort of poor man's Auto-Tune, and can also be persuaded to perform subtle Cher-like robotic glitches at extreme settings.
Two new plug-ins appear in the Master Section under a new ASIO heading, and only function with ASIO drivers. Audio Input replaces the previous 'Live Input' function, letting you capture any incoming signal from your audio interface hardware with Master Section effects, but is far more versatile since the plug-in window lets you select any number of input channels from one to eight, and your choice of sample rate. It must always be loaded in the topmost effect slot. External Gear can only be used once in the chain, and as its name suggests, it lets you plumb in hardware effects to spare ASIO input/output pairs. The latency of the send/return loop can be automatically compensated for, just as in the latest versions of Cubase and Nuendo.
The Master Section itself also benefits from a few new options. You can now store its current set of plug-ins with an audio file, its display now indicates both 'plug-ins active' (green) and 'global bypass' (red) states, and the plug-ins organiser now recognises folders. However, for me the most useful new function is Smart Bypass, which compensates for any change in loudness when processing a file, so that you can A/B switch between the original and processed versions and hear the change in sound quality rather than the change in level. The stand-alone Harbal application does this as well, and it's an extremely useful tool when making mastering decisions.
Those who aren't yet Wavelab users might consider £470 a lot of money for an audio editing, mastering, restoration and CD/DVD burning application, but considering that Wavelab 1.6 — the first version I reviewed, back in SOS October 1997 — was already £399, the price seems eminently reasonable considering the absolutely vast array of features that has been added since. I feel more comfortable working in the Wavelab environment than with any other audio application I've ever used, and rely on it a great deal for my day-to-day work. It has rarely let me down over the last nine years.
I can see a few laptop owners sticking with their existing Wavelab version to avoid the inconvenience of the new dongle, but those with desktop PCs would be fools not to upgrade, especially since existing Wavelab 5 owners can do so for just £70. Even at £130, version 3 and 4 owners should find this upgrade a bargain given the huge number of new features it adds to their arsenal of editing options.
All Steinberg recommend for running Wavelab 6 is a PC with any Pentium 4, Athlon or Opteron processor of 2.4GHz or faster, running Windows 2000 or XP, and 1GB of RAM. Unlike multitrack applications where lots of simultaneous plug-ins and soft synths are likely to be required, this seems entirely reasonable, and even the quoted minimum requirements of a Pentium III/Athlon 800MHz processor and 256MB of RAM seem feasible to me as long as you don't attempt complex plug-in chains.