Serum’s slick interface belies an extremely flexible wavetable synthesizer.
Serum is the first synthesizer from Xfer Records, creators of the enduringly useful LFO Tool. Its aims are simple: to be a ‘dream synth’, which in this case translates to a wavetable synthesizer producing high-quality sound from a ‘workflow-oriented’ interface.
Wavetables were first developed by Wolfgang Palm of PPG, the concept later taken up by Waldorf and Access (amongst others). The distinctive sound is derived from groups of digital waveforms, known collectively as wavetables. Movement and tonal complexity are introduced by scanning the table, either manually or by modulators such as LFOs and envelopes. Perhaps because of the potential for complexity, it’s a synthesis type well-suited to the graphical world of VST-land, hence the many examples that exist, trumping the older hardware in fidelity and in the number of wavetables.
Available in VST, AAX and AU formats (both 32- and 64-bit), Serum is much deeper than its unencumbered panel implies. It ships with a large vault of prepared wavetables and an extensive toolkit to roll and shape your own. As dreams go, it’s a good start...
After a brief and pain-free installation, you’re presented with an interface of unparalleled directness and welcoming clarity. Given the number of messy, confusing and inconsistent synth panels I find myself juggling each day, this one is a delight. I couldn’t even find fault with the grey background as it perfectly highlights the subtly coloured wavetables, filters, envelopes and LFOs. I doubt anyone will seek the manual with any great urgency.
Aided by a crack team of patch designers, Serum ships with approximately 450 factory sounds. A quick perusal reveals a wealth of sweet pads and bright, fuzzy strings, as you’d probably expect. Equally impressive, though, are the hard, deep and cutting basses, plus more snappy ‘plucks’ than a harpist on speed. Strolling through the collection, you’re left with an impression of shimmering uppers and rock-solid bottoms, peppered with razor-sharp leads. I did find it odd that the obvious capacity for atmospherics and cinematic sound design wasn’t more fully explored, whilst simultaneously relishing doing so myself.
The four possible sound sources consist of two independent wavetable oscillators, a sub-oscillator and a deluxe noise generator. It’s an architecture as familiar as it is logical, which means there’s nothing to stop you plunging in right away.
Selecting a single oscillator unprocessed by filter or effects, I began auditioning the factory wavetables. They’re selected from categories such as Analog, Digital, Spectral, User and Vowel, and while the names give a good idea what to expect, it’s the pervasive clarity that hits you. With over 140 tables to choose from, I could have happily spent days selecting wavetables and manually hurtling through them using the ‘WT Pos’ knob. Instead, I set an LFO to automatically update the position, achieving the task by simply dragging the header tile of a chosen LFO to the knob in question. This is one of several methods of assigning modulation and took around a second from thought to execution. The modulation range is shown by a blue arc, which in this case I extended around the knob in order to scan the whole table. You can see how many destinations each modulator has by numbers on the header tiles.
A long-time fan of wavetable synthesis, I was eager to make an original table. In Serum, a wavetable consists of up to 256 single-cycle waveforms, known as frames. Although it’s straightforward to edit an existing table, I already felt confident enough to start from scratch. A click of the pencil icon in the graphical window of either oscillator opens the wavetable editor. This is a fully stocked carpenter’s shop offering table construction by means of imported audio, drawn waveforms or mathematical formulae. The drawing of waveforms is one of those arts that seems incredibly exciting — until you try it. Fortunately, there are far superior ways to populate a range of frames. Serum can dynamically interpolate (ie. fill in the gaps) between waves using either crossfading or spectral morphing. It’s therefore entirely under your control whether the final table has smooth, buzzy or glitchy transitions.
If you opt to import a WAV, either into the current frame or to fill the whole table, you shouldn’t expect regular sample playback, although it is possible for the resynthesis process to create material that’s recognisable — speech and drum loops are great fun to experiment with. Anyone who’s spent unhealthy hours in near-darkness probing Waldorf’s ‘19-20’ wavetable will testify that a lot can be achieved with a minimal assortment of syllables, and Serum can cram in whole conversations! The maximum table size weighs in at around 2MB but most are much smaller. If you import data from large WAVs, it is truncated when the table is built.
The manual features tutorials aimed at creating high-quality wavetables from the output of other soft synths. However you choose to do it, your creations (or favourites from the factory tables) can be exported for use externally. Initially the supported formats were 8-bit (aimed at certain hardware modulars, eg. Wiard) or 32-bit WAV. Cheekily, I requested another option and was quietly stunned to see an update posted on Xfer’s forum soon afterwards. Serum now includes 16-bit WAV export too, which by happy coincidence is perfect to feed the PCM oscillators of my Roland V-Synth. My blagging wasn’t entirely selfish because plenty of other instruments can benefit from exotic digital waveforms too.
Remaining within Serum, there’s a huge number of ways that waveforms can be manipulated. This powerful magic is stored in the lengthy Warp menu. Warp’s modes and parameters are unique to each oscillator and the results can sound like a mega-sophisticated tone control one moment, a psychedelic waveform twister, wrapper and bender the next. The less adventurous may prefer to start with a familiar process such as PWM, but even this isn’t fixated on squarewaves — it’ll squeeze anything you throw at it. At the other end of the adventure spectrum, user-customisable ‘wave remaps’ are probably the pinnacle of advanced waveform squishing. Whatever Warp mode you choose, each is a great candidate for modulation, from sources located a few pixels away.
Continuing to cherry-pick through the Warp menu, there are various types of sweet, clean-sounding sync and a quantise option that turns everything decidedly nasty and rate-reduced. After that it’s extremes all the way, courtesy of AM, FM and RM (Ring Mod). All these modes force the otherwise pristine oscillators beyond their usual comfort zone.
With wide-spread waveform mashing taking place, the best way to grasp the effect of each Warp mode is to switch to the 2D view. With a single click, the scrolling wavetable becomes an invaluable real-time waveform display.
Last but not least, voices may be stacked in a powerful unison mode. The extra voices can be detuned for extra lushness, but the freakiest stuff comes with the introduction of Warp or wavetable position offsets. This makes each unison voice unique, although with a corresponding CPU cost.
The sub oscillator is basic (and no less useful for being so), but the noise generator is a bit special. It’s a stereo sample player loaded, as you’d hope, with noise of every colour. But it doesn’t stop there. Joining the noise are sampled ambient whooshes, windchimes, vinyl crackles, dripping water and atmospherics galore. Adding samples of your own is as easy as dropping them into Serum’s ‘Noises’ folder, remembering that stereo samples consume twice as many voices. This could be an issue if you’re prone to extravagant CPU-taxing chords.
Peering into another nested menu, I see four categories that ultimately hold around 90 single and combined filters. Tick boxes are provided to determine which sound sources are processed and a mix control sets the filter’s overall effect, from none to maximum. Even though Serum lacks the serial/parallel routing seen elsewhere, it’s pretty comprehensive and covers analogue-style multimode filters through to formant filters dripping with vowel-like enunciation. The quality is high throughout. Notable amongst the throng is the terrific ‘zero delay’ German LP filter and the distorted, dirty French LP (with added ‘boeuf’). It’s not just regular filters either, there are comb filters, flangers and phasers, all of which are relatively predictable, followed by ring modulation and a sample and hold filter, which are much less so. Actually, the latter reminded me of one of the weirder filters from Waldorf’s Microwave 2, which is no bad thing. Finally, you’re able to place any filter across the whole synth’s output as the filter collection is mirrored in the effects rack for convenience.
The effects rack contains 10 modules, their order reassignable with a swift mouse drag. There are no duffers here and my personal favourites were the warm and spacious plate reverb and dense, classy chorus. Most parameters can be modulated and by the same sources as the rest of the synth. Thus, if the mood takes you, you can tie the phaser speed and delay time together, or simply put both under external MIDI control.
Hyper/Dimension is a dual effect that pairs a simulated (but CPU-friendly) unison with four subtly modulated delay lines. I mention it only because it has so far found its way into every pad I’ve made.
Before moving on, distortion also deserves a quick mention. Not usually an effect I use extensively, this one has over a dozen modes and at times is so full of presence it’s almost ‘alive’. Best of all is its highly customisable waveshaper mode, complete with dual waveform editors that facilitate an almost ludicrous level of fine-tuning.
Serum is an oddly named synth built around refreshingly clean, low-aliasing wavetables. Even when forced to be filthy, it somehow retains a refined and polished character, which is no mean feat. With its zappy envelopes, highly malleable LFOs and extensive Warp tools, Serum is not just a source of shifting pads and atmospherics: it’s a very capable and versatile synthesizer. It also happens to possess one of the most elegant interfaces a complex synth was ever given.
I don’t have many wishes for future versions. An arpeggiator or a step sequencer would be welcome, but more slots in the modulation matrix would be even better. If it were possible for patches to load just a fraction quicker that would be lovely, but there really isn’t much to find fault with. I expect some will demand lower-quality or ‘draft’ modes for those ultra-clean but needy wavetable oscillators. Personally, even though my system occasionally creaked under the strain, I’ve got hooked on the quality and would rather reduce polyphony than give it up. If you have any interest at all in wavetable synthesis, grab the demo today.
Whichever method you prefer for adding modulation, Serum’s 16-slot modulation matrix shows every source and destination connection. You aren’t confined to a linear correlation between them either: the response curve can be individually shaped for each modulator. Furthermore, auxiliary modulation sources can be added to each slot, which is ideal when you need to multiply the actions of two sources. One obvious example of this being the control of vibrato depth via the mod wheel. If I’ve one complaint (and I do), it’s that those 16 slots fill up way too quickly!
With three envelopes, four LFOs, external modulators, random numbers and two sources of chaos, you shouldn’t run out of modulation sources in a hurry. Four macros shown on the front panel serve as handy performance controls, even more so when assigned to favourite knobs on a MIDI controller.
There isn’t much to say about the envelopes, except that they are fast, responsive and behave impeccably. Lovers of snappy sequencer parts will relish the ability to tweak the curves of each envelope stage. I certainly did!
The LFOs, with their variable resolution grid, should look familiar to LFO Tool users. The grid is invaluable for drawing LFOs that are intended for rhythmic or chromatic purposes. At top speed, the LFOs max out at a lively 100Hz or they can be optionally sync’ed to song tempo. The slowest sync’ed rate is a staggering 32 bars, which presents all kinds of opportunities for building evolving patches. I’ve used a number of LFO designers in the past but this one ticks all the boxes. Those of you who’ve noticed there’s no arpeggiator or sequencer can take some comfort from the LFOs.