Roland V-Synth XT: Part 2 VC1 & VC2 Cards

Photo: Mike Cameron

Part 2: We conclude our review of Roland's fabulous new rackmount synth, and look at the built-in VC1 and VC2 cards, which convert the V-Synth XT into a fully functioning D50 and a powerful vocal-processing synth respectively.

Last month, I took delivery of the V-Synth XT, and discovered that its amazing sound engine is only one third of its story. This is because, like the original keyboard V-Synth, the XT has the ability to host cards that change it into something completely different — or, in this case, two different things. On the V-Synth XT, however, two such cards are pre-installed internally, without occupying the PC Card slot on the front panel, as they do on the original V-Synth. Not merely extensions of the existing synth engine, these cards replace the 'brain' of the instrument, providing utterly different programming idioms, sounds and characters.

The first of these cards, launched earlier this year and reviewed in Sound On Sound May 2005, running under a keyboard V-Synth, is the VC1 'D50' card. The second is the VC2 'Vocal Designer' card. So let's press the V-Card button, stab the 'VC1 D50' panel that appears on the XT's touch-sensitive colour screen, and wait for the D50 implementation to load up.

The VC1 'D50' Card

In the diagram below, I've illustrated the test setup that I used for this review. As you can see, my original D50 synth provided both MIDI control over the XT and the opportunity to test the VC1's sonic authenticity.

I loaded the original factory sound set on both synths, directed their outputs to two stereo inputs on my mixer, matched the gains and defeated the EQ, and then punched the channel mutes as required. The results were unambiguous. To my ears, the two instruments sounded identical, or so close to identical as makes absolutely no difference. I tried all the classic D50 patches — 'Fantasia', 'DigitalNativeDance', 'Soundtrack', 'Glass Voices', 'Nightmare', 'IntruderFX' and so on — sounds that exemplify the original four-partial form of LA Synthesis, and which have never been accurately recreated on any other synth. When tested 'blind', I was unable to tell when a patch was played on the D50, and when on the XT.

A diagramatic representation of the equipment setup used for this review.Photo: Mike Cameron

Checking the sounds in an oscilloscope, I made some further interesting findings. Firstly, the initial square and sawtooth waves I generated with the VC1 were very close to ideal square and sawtooth waves (unlike those generated by most analogue synths). I noticed something strange about the sawtooth wave, though — its polarity inverted with every successive note I played. To my surprise, when I compared this to my original D50, I found that the polarity of its sawtooth would also suddenly switch unexpectedly while playing a series of notes. After 18 years of happy D50 ownership, this was news to me!

However, it seemed that despite this extremely faithful rendition of the D50's behaviour, something else wasn't right. When viewed through the same scope, the D50's square and sawtooth waves looked quite different to the VC1's close-to-ideal traces; the 'top' of the square wave sloped down considerably, and both waves had prominent initial transients which departed from ideal square and sawtooth traces. But then I realised that these oddities were hallmarks of the early and, by modern standards, low-quality D-A converters used in most digital synthesizers in the 1980s. Suddenly, I had it. If you read Paul Nagle's original review of the VC1 card, you may recall that Roland furnished it with both a standard mode, which transmits the D50 sounds through the modern output stage of the V-Synth unaltered, and also a 'D50' mode, designed to emulate the comparatively low-quality late-'80s output stage of the original D50. When I flipped the VC1 in the review V-Synth XT into this mode, the output square and sawtooth waves changed; the tell-tale transients appeared, and the 'top' of the square wave acquired a distinct slope down. The waveforms from the D50 and the VC1 were now identical. Clearly, the near-perfect waves I had initially obtained from the VC1 were a consequence of it being in 'V-Synth' rather than 'D50' mode. It would be hard for me to fault this emulation.

VC1 Extras & Problems

Selecting the VC1 immediately disables all the XT's V-Synth functions, so you can't (for example) play a D50 patch through the V-Synth XT's COSM processors or output effects. Nonetheless, Roland haven't left the original D50 specification entirely untouched. As well as including all of Roland's sound sets for the D50 in the VC1 (the original factory patches and the sounds from the four original library expansion cards), there are 28 new PCM waveforms, a new block of patches that use these new PCMs (which are therefore unique to the VC1), and eight blocks of RAM for your own patches. You may initially find the new PCMs rather too 'finished' for your taste; they are useable without any filtering or contouring. Furthermore, you could be forgiven for thinking that there's little point in adding electric piano PCMs and sound-effects waveforms when all the power of the V-Synth is but a single button-push away.Entering VC1 mode.Photo: Mike Cameron

Nevertheless, when I started to program patches using PCMs 101 to 128, I discovered firstly that the results retain the recognisable D50 character, and secondly, that the Epiano samples are great, as are (for example) the new Sync, Lo3Saw, FatSaw and FatSqr waves. These make it possible for the VC1 to do some of the things that you always wished the D50 would. To experience this for yourself, try P2-14, 'Next Stop Nirvana', which uses the looped PCMs 123, 124 and 126 to generate a sound in the spirit of the original 'DigitalNativeDance' or 'IntruderFX'; and P2-36, 'Long Dream', which uses 120 and 123 to create yet another long, evolving pad. In contrast, there's P2-13, 'Soft EPicenter', which makes use of PCMs 101 and 102 to generate one of a number of high-quality electric piano patches, and P2-16, 'Hard Whoover', which uses four partials based on the FatSaw PCM (number 115) to create a very contemporary bass patch.

The relative paucity of knobs on the XT means that you have little real-time control over a VC1 patch. You can assign the eight 'E' knobs to your choice from a list of 13 patch parameters, 24 tone parameters, or 54 partial parameters, but that's about all. Real editing is accomplished by pressing the on-screen buttons, and then stabbing at the physical Inc or Dec buttons, as on the original V-Synth. You can also use the Value knob (which wasn't present on the D50), but not the numeric keypad, which is a great shame. But there's another unexpected bonus with every VC1...

I was contemplating Roland's claims that the VC1 and D50 are identical when a thought occurred to me. I went and unwrapped my rarely used PG1000 hardware D50 programmer, connected its MIDI output and input to the sockets on the back of the XT and... Blimey! It worked. As far as I could tell, the VC1 responds to the D50's dedicated programmer as the D50 itself does, partial by partial, tone by tone, and patch by patch. You don't even have to be viewing the appropriate page for the PG1000 to be able to do its stuff; just move a slider, and listen to the sound change. Inevitably, there are a couple of inconsistencies; for example, the PCM select slider on the PG1000 travels from 01 to 100, so the new samples in locations 101 to 128 are unobtainable, but I'm nonetheless amazed by this degree of compatibility.

The VC1 recreates the streamlined D50 user interface perfectly.Photo: Mike Cameron

Despite all the good stuff, I did uncover a few problems during my time with the VC1. Firstly, don't follow the instructions in the manual regarding Bulk Dumping from the VC1 to a D50: they don't work. Instead, select a patch in the VC1 block you wish to dump, and then request the dump from the D50. Once it's complete, the D50 will appear to lock up, so press Exit. The screen will say 'Cancelled' but you'll find that all 64 patches have arrived at their destinations nevertheless. Secondly, dumping patches from the D50 to the VC1 (ie. in the other direction) also fails when you follow the instructions in the manual. I found that you have to add a second MIDI cable so that the two can talk bi-directionally. In both cases, it took a number of attempts to work out what was happening, with garbage ensuing until I discovered the correct combination of key presses. This is very annoying, but could be fixed by nothing more arduous than correcting the manual. In the meantime, I predict more than a few phone calls to Roland's customer service hotline!

Next, the patch numbers on the D50 and those selected on the VC1 did not always tally with one another when I used the D50 to select VC1 patches over MIDI. It was almost as if some sort of remapping was going on: patch 11 on the D50 selected patch 62 on the VC1, patch 12 selected patch 85... and so on. Happily, this problem went away when I loaded one of the VC1's memory blocks into the D50. Unhappily, it reappeared when I reloaded the problematic bank into the D50. I have no idea what was happening here.

Moving on, the VC1 cannot access the XT's PC Card slot. This is frustrating because — while the onboard memories will be adequate for all but the most fanatical LA Synthesis programmer — this limitation precludes simple portability of musical projects that include D50-type sounds. Furthermore, the software editor that accompanies the VC1 card when it's used with Roland's VariOS module is not included, and the V-Synth Librarian (see last month's review) does not support the VC1. This is a disappointment, but not a huge one — and not just because you can use a PG1000 to edit the VC1, but also because the on-screen programming is so much clearer than on the original D50.

My final criticism of the VC1 concerns the on-screen representation of the D50's joystick. This works on the V-Synth and under VariOS, but for some reason does not do so on the XT.

The VC2 'Vocal Designer' Card

In the same way that the VC1 transforms the XT (and the V-Synth and the VariOS module) into a D50, the VC2 turns it into a vocal synthesizer and vocoder. There are five classes (or 'groups') of vocal processing on offer, with up to three algorithms in each group. Strangely, there's no way to access the algorithms themselves. To create a new patch using a desired algorithm, you have to select an existing patch that uses it, copy this to a new memory location, and then edit it into what you wanted in the first place. This is daft!Switching to VC2 mode takes about 12 seconds, progress being shown by the bar at the bottom of the screen.Photo: Mike Cameron

The five groups are: Modelling, Vocoder, Keyboard, Processor, and Poly Pitch Shift, all but the last of which are based on some form of Modulator/Carrier vocoding. The Keyboard algorithms use an internally generated wave to modulate an internally generated Carrier, whereas the Modelling and Vocoder algorithms use an external signal to modulate an internally generated Carrier, and the Processor algorithms use both external Modulators and Carriers. The odd one out is the Poly Pitch Shifter, which allows you to process the incoming signal, applying formants and so on, without any kind of Carrier wave. Let's start by investigating the first of the algorithms in the Keyboard group, Keyboard Choir, to illustrate how the VC2 works...

The four programming sections for this group are called Wave, Carrier, Vocoder, and Effects. The first page (called 'Osc') in the first of these presents you with a window that lets you select a small range of 'oohs' and 'aahs' as Modulators, as well as a handful of unexpected waves with memorable names such as '113 WahGtr +' and '116 WahCutG+'. If these are not what you're looking for, you can import your own waves in much the same way that I described last month when discussing the equivalent capabilities for the XT itself (see the box on the previous page for more on this).

If a sample has a lot of internal structure, it's interesting to be able to speed or slow it by sync'ing it to the VC2's internal clock. This allows you to change the rate of articulation without affecting the pitch of the resulting sound. You can click on the 'Always' button and have the sample play and loop indefinitely, and on the next two pages ('Vari' and 'TVA') you can vary the nature of the selected Wave by determining and then using MIDI controllers to modulate its pitch, its speed, the position of its formant, and its amplitude. Experimenting with these facilities can lead to all manner of new and interesting sounds, particularly when you slow a sung phrase to a fraction of its normal speed. The evolving choir that results can be magical.

Having defined your Modulator, you now need to specify the Carrier, which can comprise either one or two oscillators. Osc1 offers a selection of PCM waves, while Osc2 provides an extended set of PCMs and digitally generated 'analogue' waveforms, the latter of which offer variable pulse width, panning, and sub-oscillators. Again, there's an extensive range of parameters available, with five sub-pages that offer independent pitch control, portamento and vibrato for each oscillator, independent equalisers, 'growl' (which is tremolo), plus an extensive amplifier section with variable Attack and Release, panning and key follow. As elsewhere, you can control many of these parameters using MIDI controller messages, making this section very flexible and performance-friendly.

Next, we come to the Vocoder section, which uses the wave to modulate the Carrier, and thus generate the sound you hear. You can output three signals from this section: the vocoded sound, the so-called 'Natural Voice' (actually yet another name for the Modulator), and any unpitched sounds such as consonants and plosives that are detected in the Modulator. The most important control here is the Formant selector, which allows you to choose from 16 different vocal characters: Flat, Hard, Medium, Soft, Male, Female, Child, Husky, and so on. Experimenting with this in conjunction with any given Modulator and Carrier waves allows you to create a huge range of choral sounds. Some are classic, some are innovative, some are brash, others are subtle, but all are of excellent quality and usability.

In my view, it's almost impossible to create a naff sound using the Modelling Choir algorithm. However, there's a caveat, and the manual doesn't mention it. If you use both Carrier oscillators, and select one of the analogue-style waves in Osc2, the polyphony can drop to just three notes.

The Keyboard Vocal algorithm is very similar to the Keyboard Choir, with two major differences. Most obviously, the polyphony is even more limited, dropping as low as two notes when dual-oscillator Carriers are used. This means that it is more suitable for switching into Mono mode in the Carrier page, and using it to play solo lines such as might be sung by a classical singer. Secondly, a different selection of Carrier waves is provided, some of which help to generate 'human' timbres, while others — particularly in Osc2 — help to generate the type of synthetic, quasi-voiced timbres that programmers of large analogue synths have been trying to obtain for 40 years.

The last algorithm in the Keyboard group, Keyboard Analogue, adds nothing new, but uses a subset of the previous Carrier waves to generate a different range of polyphonic (and more obviously keyboard-like) timbres.

By and large, everything falls to hand once you have mastered what's going on. My only confusion was caused by the manual's insistence that there's a 'Tone' parameter in the Keyboard algorithms' Vocoder sections. There isn't — although it does appear in some of the other groups.

Moving on to the Modelling group, in many ways, this is the same as the Keyboard group, and it similarly offers three algorithms, logically entitled Modelling Choir, Modelling Vocal, and Modelling Analogue. The difference is, of course, that you create the Modulator by presenting vocals or any other external signal to articulate the Carrier. The main programming difference is, therefore, the replacement of the Wave section with a 'Mic Setting' section. This offers three-band pre-equalisation, a noise suppressor (Gate), and a compressor/limiter to even out the amplitude of the input before processing. You can name and store up to eight input setups to aid use with different microphones, or in different environments.

There is, however, an important addition in the Vocoder section: a switch catchily entitled AutoNoteSw. If this is on, not only is the Modulator generated from the input signal, but pitch information is derived from the input and sent to the Carrier, so that you can produce melodies without providing MIDI Note information. As you would expect, the pitch-to-note converter is monophonic and, unfortunately, you have to sing well (or, if using an instrument to provide the input signal, play very cleanly) to generate acceptable results. You can override the pitch identified by AutoNoteSw by supplying MIDI note data, whereupon these, rather than AutoNoteSw, will determine the output pitch(es). Stop playing, however, and the pitch will jump back to that determined by AutoNoteSw. When used 'live', you can use this to provide accompaniments to the Natural Voice (ie. Modulator) signal. I can also envisage times when you might use this in the studio to correct poor pitching in a vocoded passage.

There are also three algorithms in the Vocoder Group: Vocoder Choir, Vocoder Solo and Vocoder Vintage. They all produce typically 1970s-style vocoder sounds, but with a clarity that would have been almost unimaginable 30 years ago, making these algorithms perfect for recreating the appropriate bits from (say) Pink Floyd's Animals. The controls are very similar to those of the Modelling group, but the Formant setting in the vocoder section has disappeared, so the nature of the output cannot be modified in this way.

It's worth noting that it's in the Modelling and Vocoder groups that the ability to add unpitched elements back into the vocoded signal yields its greatest benefit; doing so can greatly increase the intelligibility of vocoded sounds.

This VC2 patch uses the Vocoder Vintage algorithm to emulate Roland's classic late-1970s VC330 vocoder.Photo: Mike Cameron

Next, there's the Processor group. Just as Roland have replaced the internal Waves to create the Modelling and Vocoder algorithms, they've provided two algorithms that also eliminate the internal Carriers, replacing these with any signal you present to the Left input socket on the rear panel. These allow you to modify and vocode any Carrier of your choice; synthesized sounds, drum loops, your favourite Spice Girls CD... or whatever. The Type 1 algorithm produces a clean, modern sound, while Type 2 creates something more akin to what you would have obtained using the External Synth input on a vintage vocoder such as a Roland VP330. The possibilities are enormous, although the results can be of variable quality because the VC2 is very sensitive to the input levels of both the Modulator and Carrier signals.

Finally, we come to the last of 12 algorithms, the pitch-shifter. This takes the signal presented to the microphone input and shifts it to the pitch (or pitches) specified by your MIDI controller. It works reasonably well, with only moderate glitching if the signal that you present is pure and monophonic, and it seems to work best if the signal is monotonal as well as monophonic. But woe betide you if you present a polyphonic signal and try to shift this to multiple pitches — the result is cacophony!

Although the Mic Settings in the pitch-shifter are the same as elsewhere, the algorithm has no Carrier or Vocoder sections, both of which are replaced by a 'Poly Pitch Shifter' section. Like a Carrier, this offers two independently tuneable oscillators that allow you to impose your choice from 11 formant types on the incoming signal. You can then determine how much of the input, and (separately) how much of the unvoiced signal components in the input are passed, and — as elsewhere — apply glide, vibrato, EQ, a basic envelope, and so on.

You may ask how this differs from conventional vocoding, and the answer is not by much. Given that the shifter's oscillators allow you to choose independent formant styles for each, and to shift the pitches of these to create all manner of new timbres, the distinction is — as far as the user is concerned — mostly one of implementation. Having said that, it would be nice to know exactly what's going in all of these algorithms, but Roland's documentation is silent on this point.

The Effects

Whichever algorithm you use, the effects at the end of the signal chain remain unchanged. These echo (no pun intended) those of the V-Synth engine, with the full set of V-Synth MFX and Chorus algorithms. However, the routing in the VC2 is much more limited, with just an input to the MFX module. You can still direct the signal to the chorus and reverb once it has passed through the MFX, but you cannot route it directly to either from the TVA (see the screenshot below). I have no idea why Roland have imposed this limitation, and I can think of no technical reason why it should be so. Furthermore, the VC2 does not have the 10 reverbs of the V-Synth's engine — it has 14, losing the Plate, Non-linear and Delay options of the V-Synth, but gaining seven stereo algorithms, ranging from Stereo Plate to Stereo Cathedral. This may seem strange, but the new set is focused more closely on vocal processing, and is more appropriate than the XT's reverbs would have been.

The V-Synth XT's effects, shown here being accessed in VC2 mode (which is why the routing from the TVA here has to pass through the MFX block at the top of the screen — see below for more on this).Photo: Mike Cameron

Pluses & Minuses

In contrast to the VC1, the VC2 appears to be a complicated beastie. But when you strip away its multiple algorithms, formant shaping and whatnot, it's no more nor less than a modern-day equivalent of Roland's VP330 Vocoder Plus from 1979. This offered choral sounds that you could play from the keyboard, and it even had an ensemble effect that allowed you to switch from choirs to solo voices — rather like the Keyboard algorithms here, in fact. The VP330 also provided a microphone input that accepted external signals as Modulators, thus allowing you to 'vocode' your voice or other instruments, again with the use of ensemble, or not — rather like the Modelling, Vocoder and Poly Pitch Shift algorithms, in fact! The VP330 even sported a second signal input on its rear, allowing you to provide a second signal to act as a Carrier — just like the VC2's Processor algorithms!

Once you've grasped this, the VC2 becomes much simpler to use than the descriptions of the algorithms might suggest. Cut through the jargon in the less-than-perfect manual, and you have an instrument that produces great vocal sounds when played like a conventional keyboard, great vocoded sounds when you feel the urge to emulate Mr Blue Sky (which I hope is not too often), and great new sounds when you present external signals to both the Modulator and Carrier inputs. Simple, huh?

Indeed, it can be really simple to use the VC2. Select a patch that you want to use, and the 'Top' screen displays the most important controls without you ever having to delve any deeper. For example, a Vintage Vocoder patch provides sliders for the Mic input level, the Carrier level, the Vocoder and Natural Voice levels, plus the chorus and reverb output levels, as well as knobs for the Carrier release time and Vocoder tone (see the screenshot on the previous page). Nonetheless, it would be a shame to stop here, because the VC2 offers so much scope for control using pitch-bend, aftertouch, most MIDI Continuous Controller numbers, the dual D-Beams and Time Trip pad of the V-Synth keyboard, and more.

There's more good stuff that I can only mention in passing. For example, you can create an alternative (not even-tempered) scale, any patch can be set to monophonic (with a legato function) or polyphonic playback, and a Project (the highest level of memory) can contain up to 448 patches and up to 127 waves. Oh yes, and the icons and menu at the top of the er, Top screen also provide access to a chord memory function and to the tempo for synchronising the effects and waves that use tempo data (see the box on the left for more details).

But, in small ways, it soon becomes apparent that the VC2 is an add-on to the V-Synth architecture, not an integrated part of the original design. For example, various buttons — such as the Preview, Key Pad and '9' buttons — remain unused in VC2 mode, or have uses that are not the same as the front-panel legends would suggest. Likewise, using the VC2 is perhaps not as slick as using the main V-Synth engine, because you can't use the numeric keypad to enter parameter values. This means that there's lots of winding of the Value knob, which soon becomes tiresome. In the same vein, I soon discovered that the V-Synth Librarian is unable to support VC2 patches and waves, which seems a bit of an oversight given that the VC2 is part of the standard configuration of the XT. But apart from this, it's hard to criticise the VC2.

Conclusions

When I first saw the original V-Synth, I thought that it was perhaps the first keyboard synthesizer since the Korg Z1 to offer something truly new and appealing. Strangely, it has remained largely unexploited, but with the advent of version 2.00, the V-Synth concept has come of age, and with the arrival of the V-Synth XT, it has become relatively affordable.

Although the cards are priced at just £100 and £130 (respectively) to owners of the V-Synth and VariOS, I think that the 'free' inclusion of the VC1 and VC2 is a key element in the appeal of the XT. In part, this is because, 18 years after buying my first D50, I still use its sounds regularly, and in part, it's because the quality of the VC2 makes many older vocoders obsolete, freeing valuable studio space for something else.

However... because the XT can only run one 'brain' at a time, you have to treat it as three mutually exclusive instruments, much like having three rackmount synths and only one power lead. In the studio, this is not a problem, because you can lay any LA Synthesis and/or vocoded parts as audio and then sequence or play the V-Synth parts (or vice versa). On stage, however, the VC1 and VC2 are most likely irrelevant, because I can't see many players sacrificing the massive power of the XT to access them.

This aside, the XT is a superb synth, not a half-baked rehash of the V-Synth, but a well-thought-out repackaging in which issues such as the replacement of the Time Trip pad and the 'hot swapping' between V-Synth, VC1 and VC2 engines have been carefully considered and implemented.

I'm now convinced that Roland's designers got it right when they separated the V-Synth from their workstation technologies. The pressure to reduce the cost of access to the V-Synth engine by integrating the XT into the Fantom series (much as Korg did when the Z1 became, in essence, a plug-in card for the Trinity and later the Triton) must be significant. Indeed, I would be surprised if this did not happen in the future. But, for now, I like the V-Synth and the XT as separate instruments, each designed specifically to do the job that they do so well. If you're serious about synthesis or sound design, I can think of no excuse (other than a purely financial one) why you should not have an XT tucked away in your rack. 'Nuff said.