Photos: Mark Ewing
I can't believe it's more than four years since Roland's V-Synth entered my life. Its remarkable sample-warping abilities, its COSM (Composite Object Sound Modelling) processors and its innovative performance controls turned me into a gushing fanboy during my initial review for SOS. So at the mere mention of an updated version, the V-Synth GT, I began hoarding pocket money and mentally freeing up studio space...
For those readers with an imperfect grasp of their V-Synth history, the original V-Synth was an innovative instrument that combined modelled analogue waveforms and PCM variphrase oscillators with user sampling. These audio sources were routed in a semi-modular fashion through a selection of COSM processors such as standard filters, amp models, waveshapers, side-band filters, frequency shifters, comb filters and more. Features such as the twin D-Beam controllers, 'Time Trip' pad, dedicated knobs and responsive touchscreen ensured that its unique "elastic audio" (as Roland described their 'Variphrase' concept) was highly malleable in performance.
Although basic multitimbrality and keyboard splits were included, there were limitations: for example, the arpeggiator could not be confined within keyboard zones and the maximum polyphony of 24 notes was soon eaten up by complex patches. The new V-Synth GT variant features a "dual core" structure; think of this as two V-Synths in one body, since each half (or 'tone') can be treated like a complete patch of the original. This offers layering potential previously impossible — and the synth's arpeggiator can now be allocated to a specific keyboard area in split keyboard setups. Polyphony has increased too, and in most cases effectively doubles the capacity of the original.
Like the earlier XT rack, the GT has a colour touchscreen and sacrifices some of the original keyboard's dedicated knobs for a row of assignable ones. Conceived as an "expressive synth", the GT also features brand-new technology, in the form of Articulative Phrase Synthesis. APS employs a separate pool of waveforms and performance models to replicate some of the nuances of human performance.
Both the original V-Synth and its rack counterpart, the V-Synth XT, have received thorough examinations in SOS; to gain a complete perspective I therefore recommend that you read this review in conjunction with the earlier ones from the May 2003 and September/October 2005 editions of SOS.
In classic black with silver plastic end-cheeks, shiny-top knobs and a host of glowing orange buttons, the V-Synth GT makes a striking figure. When the review model arrived, it was in a box that was visibly well-travelled. Fearing the worst, I opened up and set about extracting chunks of polystyrene from keyboard and buttons. I needn't have worried; Roland's baby has a sturdy metal body and I suspect it'll take more than a few cavalier couriers to beat it into submission.
Bonus style points are earned for the new synth by the neatly recessed colour screen and Time Trip pad (an innovative controller with various uses), and by informational panels lit by splashes of blue light. The aforementioned 320 x 240 colour TFT touchscreen sits at the heart of operations. With numerous pages of on-line help, a map function and graphics galore, it does an excellent job of conveying a large amount of information, and it only occasionally feels as though there's too much crammed in. There's even a selection of screen savers and wallpaper, and you can import your own bitmaps to do the same job, too, if such things turn you on. For the most part, I found the interface a delight, apart from several occasions when I noticed the screen laboriously redrawing itself. This typically occurred when complex patches were involved, or if I tried to simultaneously tweak and perform. Assuming the CPU shares its activities between the synth engine and screen updates, perhaps some optimisation can be achieved in a future operating system.
Compared to the first V-Synth, there are fewer knobs in dedicated roles, although the most important controls for oscillators, COSM processors and envelopes are retained. I missed having instant access to oscillator levels and LFO depths but, as I mentioned earlier, by way of compensation the GT offers eight assignable knobs (labelled E1 to E8), positioned below the screen. As far as possible, these correspond to on-screen objects (as they did on the V-Synth XT) and assist greatly during editing. In play mode they are assignable, but sadly on a global rather than a per-patch basis. However, for some bizarre reason they can duplicate the functions of the GT's existing knobs and sliders, yet can't be assigned to mimic all those of the original.
On balance, the V-Synth's new look gets a positive thumbs-up, whether for no-brainers such as the dedicated Write button (where previously there was merely an option in a menu) or for the more practically useful enhancements, including one that may not be so obvious: the D-Beam infra-red light controller has been moved away from the centre of the synth by 13cms. This minor feat of engineering is more important than it may at first appear, because it offers the user a fighting chance of stacking a smaller keyboard above the V-Synth GT without permanently triggering the D-Beam's, er, beams. Given the opportunity, I would have moved the D-Beam even further, but this is definitely progress.
Incidentally, the GT's knobs and sliders transmit System Exclusive data rather than CCs (Continuous Controllers), so don't expect to control your soft synths with them. And, in common with its older sibling, the V-Synth GT offers few concessions for those hoping to use it as a master keyboard. Having been reinvented, its remit now seems to be that of a stand-alone performance instrument, and in accordance with this it has also dropped the older models' multitimbral mode and rhythm kits. There isn't even an option to control its two layers via separate MIDI channels.
If we peer over the back of the synth for a moment, we'll see further evidence of change, although the audio output count remains the same, at a stereo pair for the combined, effected output, plus an assignable, untreated pair. Newly arrived is a microphone socket, complete with phantom power; this, as we'll discover, is a pre-requisite for one of the coolest GT features (see the 'Vocal Designer: Vocoding & More' box). And, as you'd expect, digital I/O has been preserved in either optical or coaxial format.
USB connectivity is present too, which is just as well, as it's increasingly becoming the norm. Two USB sockets are provided, one of which mirrors the USB audio, data and MIDI capabilities of the V-Synth XT (the original keyboard lacked USB audio functionality). The second USB port, compatible with USB 2.0 backup devices, is concerned with data handling. This supersedes the familiar PC Card slot, and the switch in media types has implications for existing V-Synth owners contemplating the GT. For a start, it means we can't use the backup cards we already have, but more seriously it denies continuity of the VC1, that cunning optional extra responsible for temporarily turning the V-Synth into a D50. The card slot was a snug, safe and, above all, internal receptacle, and I personally feel it's a step backwards to replace it with a protruding USB memory stick.
A numeric keypad offers a direct means of patch selection. It is accompanied by several other methods, including category or text string searches, to help locate any patch you can think of. A major annoyance for anyone wishing to browse through a list of patches whilst playing is this: as the selection appears on screen, one of the patches in the list will always be automatically selected, replacing the one you're playing. Fortunately, for live work, there is a handy Patch Palette into which 64 of your favourites can be saved for instant recall. When we first encountered this feature you could browse the entries in the palette prior to making a selection — but in the V-Synth GT this feature is broken. You can still store eight banks of eight patches, as before, but there is no way to view their names afterwards. Nor can you see the contents of a palette location that you are about to overwrite!
What's Articulative Phrase Synthesis?
Articulative Phrase Synthesis is the component of the V-Synth GT that is completely new. Described as neither sampling nor modelling as we currently understand it, APS simulates the way an instrument responds to human playing techniques. To accompany this revolutionary means of expression, two new switches have appeared. "Hold on a moment," I hear you ask, "is that all?" Well, yes, because the V-Synth already boasted some of the best performance controls to grace a synth in recent years — twin D-Beams, the Time Trip Pad, control pedal inputs, a velocity and aftertouch-sensitive keyboard, and a combined modulation and pitch stick.
Performance controls are an integral part of AP Synthesis. To grasp how it works is easy if you have the instrument at your fingertips! If not, the best way I can explain is to say that a selection of phrase models are provided, each with parameters determining their response to controllers and dynamics. Roland have chosen five instruments they consider to be most expressive. These are violin, erhu (Chinese violin), saxophone, flute and 'multifade' (a 'unique phrase model without any real-world inspiration', in case you were wondering). While you play, controllers should be used liberally to manipulate key elements of these models. The elements could be attack time, scrape level, natural feel, vibrato, portamento, breath level or whatever else is appropriate.
Powering the five performance models are a selection of waveforms — 38 in total — such as violins, saxes, pipes, brass, some synthesizer waves, and even 'robot voice', the token weirdy. There are restrictions in terms of the waveforms that can be combined with each model, but the display helpfully greys out those that are not allowed.
A large number of parameters are involved, and I saw at once how easy it would be to get bogged down in them. Fortunately, there are a number of excellent factory patches and a Getting Started manual to guide you through. Because, make no mistake, you won't get the best out of APS without practice!
Each phrase model is monophonic, with the exception of the violin, which offers four notes of polyphony. It was to the violin that I turned first, learning to vary my attacks and to produce slurs and a realistic-sounding vibrato. APS imparts far more character than you hear in a conventional violin sample, its pitch modulated via an LFO. In this model, the two switches are used to invoke tremolo and pizzicato, the latter really giving the impression you are plucking the same strings you just bowed.
Next up was the erhu, evoking haunting Crouching Tiger memories. Its tonality, portamento and vibrato were markedly different from the violin, illustrating how much of what we hear is shaped by performance. I went on to explore further examples of APS, although none tugged at the heartstrings quite so insistently. There was a selection of tenor saxes that never quite convinced (although the soprano sax was sweet) but pretty much all of the woodwinds sounded believeable.
Lastly, I checked out the Multifade model. It works in a very similar way to the others, providing control points for velocity, aftertouch and modulation, and routing them to such expressive outlets as portamento time, vibrato depth, attack and 'sub-tone level' (a kind of background noise whose character depends on the source waveform). Even when combined with some of the synthesized waveforms, I didn't rustle up anything too outlandish; the most unusual results I achieved used tone structure four, which routes APS through the COSM section. There, at last, I started to discover hints of expressive solo patches that didn't immediately bring to mind existing instruments, something a little lacking in the factory set.
With other aspects of the GT demanding my attention, my final thoughts on APS are that it throws down the challenge to consider fresh approaches to keyboard expression. Personally, I wished there was a way to push it further, to incorporate user waveforms or create new APS models, but as this is still early days, who knows what the future holds?
The V-Synth GT expands the earlier machines' polyphony from a maximum of 24 notes to 28. For once the advertising claims appear to undersell the reality, as, although the increase doesn't sound terribly significant, in my experiments with resource-hungry patches I found the polyphony to be almost double that of the original keyboard. Patches that previously would have left only four notes free appeared on the GT to offer seven or eight, which was a very pleasant surprise.
Taking over two minutes to boot is the penalty for adding powerful new technologies, including Vocal Designer (based on the much-admired VC2 card) and Articulative Phrase Synthesis. We'll discuss these soon, but for now just remember that each can be utilised without a reboot. This is a major plus compared to the 'power recycle' method of the original keyboard and even the hot-switching of the XT.
Roland say they've "integrated a new dual-core V-Synth engine" into the V-Synth GT, and due to its 'dual core' nature the GT can offer genuine keyboard splits or layers. Each may be individually marked for control via the arpeggiator, meaning that you can arpeggiate a bass line or drum loop while freely playing a solo. I've been waiting for that capability on a V-Synth since 2003! This is made possible courtesy of a dual tone arrangement for patches. Referred to as either Upper or Lower, a tone is functionally equivalent to what we previously thought of as a complete patch. In all important respects this implementation adds up to double the original's power. There is even a designated multi-effect processor for each tone. However, if you've owned earlier Roland synths such as the MKS50 or MKS70, the word 'tone' might generate a shiver of concern. Unfortunately, such fears would be justified because any tone in the pool of 896 tones can be shared by any of the 512 patches. When you edit a patch, then save its Upper or Lower tone (or both), there's no way to know how many other patches will be affected. It could be one or dozens!
If a GT patch took unique copies of each tone (as a Korg Radias patch does, for example), then all problems of this nature would vanish. A quick calculation shows that had there been a mere 128 extra tone locations, every patch could have had a unique Upper and Lower pair. I'm really hoping Roland will revisit this implementation.
Each tone is freely configurable in five different structures, three of which are equivalent to those of the original V-Synth. Structures define the paths by which oscillators are mixed, routed through dual COSM processors and finally sent to the amplifier stage. AP Synthesis (see the dedicated box elsewhere in this article) is included in parallel within every structure; it may be mixed in at the effects stage but doesn't interact with the PCM or analogue oscillators. Only in structure four does it slot in, taking the place of oscillator two. But there is no structure that allows separate COSM processing for AP Synthesis and a standard V-Synth oscillator. Finally, a fifth structure is provided that features just Vocal Designer and AP Synthesis.
Structures can be allocated freely, with only minor restrictions; for example, AP Synthesis and Vocal Designer can be active in either the Upper or Lower tone but there can only be one instance of each overall. Finally, it's worth remembering that every tone may be split into 16 keyboard zones, as before.
Vocal Designer: Vocoding & More
Vocal Designer is based on the VC2 card developed for the V-Synth, whose technology percolated through into the VP550 stand-alone keyboard (reviewed in the June 2007 edition of SOS). The VC2 was primarily tailored for producing quality vocal or choir effects, but in the V-Synth GT this represents the tiniest fingernail of its potential. As before, various different formant types (soprano, baritone, talkbox, and so on) are available, each selectable according to vocoder type. Sadly, I could find no equivalent of the VC2's monophonic pitch follower which, though a little erratic (certainly when driven by my voice), was a lot of fun. Instead, you can expect a wide selection of typical vocoder tones, from vintage VP330 emulations to highly realistic choirs. The factory patches reflect the best of those we've heard before and offer a few more wacky possibilities too. I suspect they are only scratching at the surface, though, because of the way Vocal Designer has now been neatly integrated into the whole. Here you gain the exciting prospect of incorporating any two PCM or analogue-modelled waveforms as carriers for vocoding. Furthermore, your VD patches can utilise the V-Synth's arpeggiator, multi-step modulator, COSM processors and all those MIDI-sync'ed LFOs and effects.
This gives Vocal Designer a bigger, broader personality, and merely to ponder all the options could fill an entire review. Consider briefly the advantage of using Variphrase oscillators over traditional samples. When you play chords comprised of voice or speech samples, Variphrase renders all notes of the chord perfectly in sync. This kind of trick excited us in the Roland VP9000, and when combined with a vocoder of superior quality the possibilities for freaky voice textures, harmonies and ensembles are staggering. Similarly, if you use the 'step' facility of the PCM oscillators, you can step through a sample via the keyboard (or arpeggiator) while applying vocoder intonation with the microphone. Speaking of which, Vocal Designer can memorise input level and other settings for up to eight mics, and comes with its own noise suppressors, equaliser and compressor. In fact, the only limitation I could find (if it's fair to call it that) is that the microphone input is hardwired as the vocoder's modulator. I'd have liked to internally route, say, a PCM oscillator as modulator while another waveform served as carrier.
Of the 512 patch locations, over 200 are initially empty, and those presets that are provided are of mixed quality. For instance, there are rather too many indifferent one-finger wonders, drenched in effects and arpeggiation. These accompany a selection of warm, wide pads, scarily realistic violins and flutes, classic vocoder simulations, and pearls plucked from the earlier models. The V-Synth GT even resurrects a few of the PCM multisamples that were discarded in version 2.0 of the V-Synth OS. This facilitates the return of some acoustic piano patches. While not on a par with those of the average workstation (their waveforms are sourced from the JD800 synth), I nevertheless greeted them like old friends.
The V-Synth GT gains 14MB of extra wave memory over its predecessor. Even though the V-Synth is not a sampler in the conventional sense, I'd like to have seen more than the 64MB that this adds up to. I know comparisons between instruments don't always make sense, but when another Roland synth (the Juno G) can address over 500MB of sample memory, a chap can't help but feel a tad disappointed. Not that I'm knocking the additional memory — over three minutes of extra sampling is no bad thing — but I guess the whole 'dual core' blurb boosted my expectations.
In addition to wave memory, internal flash memory has been increased from the old synth's 10MB to a much more healthy 50MB. This can serve as a receptacle for your projects and samples, so it's a pity there isn't enough to house a complete 64MB memory-maxed project. If there was, you could avoid booting from an external USB memory stick.
Other than these memory changes, sampling is almost identical to previous versions. The 'almost' is because I discovered that, in the current OS, the sample import facility has gone AWOL! Fortunately, there is a workaround. By copying your WAV files (16-bit, 44.1kHz) into the wave folder of a V-Synth GT project you can then edit, encode and finally save them. Although this is a manual operation and you must rename your WAVs according to the convention used in projects (ie. prefixed by a number), it could be sufficient until Roland get this sorted!
Soundshaper 2 is a simplified form of editing that hides many of the deeper parameters, presenting only those deemed most important. It's designed for painting in broad strokes and greatly reduces the number of screens through which to navigate. At any time you wish, you can switch into 'Pro Edit' to access everything.
Given the ease of use offered by the touchscreen and by a goodly selection of knobs, you'd hardly think Soundshaper 2 was necessary. Personally, I didn't use it much in its original incarnation and probably wouldn't here either — especially since this version lacks the fabbest feature of the former: an on-screen X-Y axis with which you could draw in combinations of parameters with a finger.
The effects section has been improved by the addition of a second multi-effect, thus giving one each to the Upper and Lower tones. Other than the fact that multi-effects are now known as 'tone effects', that's really about all there is to say. The 41 multi-effect algorithms are all as before and include a wealth of delays, phasers, distortion and so on.
Reverb and chorus are common to both tones (and to APS) and are largely unchanged from previous models too. There are three small additions, in the form of extra reverb algorithms that were plucked from Vocal Designer (see 'Vocal Designer' box).
When Roland created OS version 2.0 for the original V-Synth, they upset a lot of owners by making existing projects incompatible. I personally wrote off a heap of good stuff, and so was justifiably anxious to learn whether, this time, things would be better. Well, there's good news and bad news. The good news is that a project import option is present and it works perfectly. Projects are loaded and all 512 patches are successfully converted to GT patches, along with corresponding tones. As for the bad news... well, I'm sorry, but as of today there is no means of importing individual patches from other projects. So having restored your older work, you can't then include any of the new vocoder, APS or other material from the GT factory set. Due to architectural changes, there's no means of importing patches created for the VC2 card either.
I find this a difficult conclusion, as I'm torn between my love of the V-Synth's 'elastic audio' concept and my frustration that its replacement takes several steps in a very different direction. My own V-Synth is the star of everything I do — live or in the studio — so the prospect of having two such engines in a single keyboard, plus enhanced polyphony, a zoneable arpeggiator, a wondrous vocoder and an innovative means of introducing expression, should have been a dream come true. The dream faded slightly on discovery of the GT's shared tone pool — I loathed this approach 20 years ago and personally can't contemplate dealing with it again. Your mileage may vary.
In the interests of balance, I should point out that most of the plus points from earlier reviews still apply. As for the shortcomings I've listed, Roland are aware of all of them, plus some that I don't have space to document here. I think we can safely assume there will be solutions in a future OS — at least for the blatantly broken or missing features, such as viewing the Patch Palette, and the sample and patch import.
The V-Synth GT is a serious instrument with a price tag that reflects this; its colour display is the doorway to a bewildering array of possibilities both old and new. Of the latter, the Vocal Designer implementation is superb. As it has always pained me to sacrifice Variphrase in order to use my VC2 card, the prospect of having Vocal Designer alongside my regular patches is quite a temptation. Only AP Synthesis still feels like a technology seeking its rightful place. Whilst it doubtless has the potential to take emulation of strings and woodwind to a whole new level, I felt Roland's choice of expressive instruments was simply too unadventurous — especially in a synth formerly considered an antidote to convention. Maybe my impressions are tainted because I was force-fed too much Hovis as a kid and can therefore never appreciate the subtle nuances of the oboe! I won't deny that I thoroughly enjoyed playing with APS, and if it marks the start of an era where manufacturers genuinely put synthesizer expressiveness to the fore, the V-Synth GT could prove to be an important milestone.
V-Synth GT At A Glance
Dual-core (two tones per patch, plus splits and layers).
Separate multi-effect for each tone; common reverb and chorus.
512 patches, 896 tones.
28-voice maximum polyphony (depends on resources used).
64MB sample RAM (non-expandable).
49.5MB on-board flash RAM storage.
External data storage: optional USB 2.0 flash memory.
Arpeggiator (user programmable, can record control changes).
Articulative Phrase Synthesis: 38 preset source waveforms; five phrase models.
Vocal Designer (available without reboot).
Colour TFT touchscreen.
Version reviewed: 1.02.