Steinberg's VST Instrument technology allows software synths to be fully integrated into any VST 2.0-compatible host software, such as Cubase VST. The first major releases to arrive are both recreations of established analogue classics. Martin Walker tries them out.
For anyone who wants the sounds of vintage analogue synths without the problems of keeping them in tune, getting them serviced as parts become ever more scarce, and retrofitting MIDI to adapt them to the modern studio, recreating them in software seems an ideal solution. The problem in the past, however, has always been that software synths rarely sounded anything like the original article. Running them alongside your MIDI + Audio sequencer could also throw up a host of difficulties. The former problem is being resolved, thanks to faster processors, which provide enough power to model the sounds more realistically, and talented software developers, who understand what made those original synths sound the way they did. The latter has been tackled by Steinberg with their VST 2.0 plug-in technology, which provides an ideal integrated environment, as well as much tighter timing than is normally provided by a MIDI instrument (see the 'Timing & Latency' box on page 106).
This integration lets you activate any VST synth when you are already working on a song, apply any of the facilities of the VST mixer such as EQ or insert and send effects to each of its sounds, an
NATIVE INSTRUMENTS PRO FIVE £149
Diverse range of sounds.
Up to 32 voices from each Pro Five launched.
Supplied with a huge bank of 512 sounds to get you started.
Fat unison mode available.
LFO frequency can be synchronised to tempo.
Some people may not like the tiny controls.
Locating a particular program among 512 can be tedious.
More use could have been made of the stereo output.
Pro Five is a well-implemented VST Instrument that is capable of a surprising range of quality sounds from rich and lush to hard and metallic, with far greater polyphony than the original design.
Steinberg Model E
The Minimoog Model D is probably the most famous and influential synth ever designed. With its three oscillators, fat filter, fast envelopes, and versatile routing (for its time, anyway), it set the standard for compact, prepatched synth design. Steinberg's virtual copy employs the same routing and adds polyphonic capability and MIDI support. The VCO section contains three oscillators, each switchable over six octaves and providing a choice of six waveforms. Oscillators 1 and 2 have triangle, triangle/sawtooth mix, sawtooth, and three widths of pulse wave from square to narrow rectangular, while Oscillator 3 swaps the triangle/sawtooth for reverse sawtooth. Oscillators 2 and 3 also have fine frequency controls over a range of ± seven semitones.
All three oscillators, along with a choice of white or pink noise, can be combined in the Mix section using virtual slider controls (a departure from the rotaries of the original). However, the external input of the original has gone, so sadly you can't route other audio tracks through the filter stage, or feed the output of the synth back into the mixer for those overdriven guitar-like solos.
The output of the mixer is fed into the filter, which has Cutoff (frequency) and Emphasis ('Q' or resonance) knobs, and an Amount control to modulate the filter frequency via an ADS envelope generator. Unlike the original Minimoog, the filter can be switched between the classic 4-pole type and an alternative 2-pole mode. The 4-pole mode sounds fatter as it adds subtle distortion, and will self-oscillate if the resonance is set high enough, while the 2-pole mode sounds grittier and more 'digital', but takes less processor overhead. As in the original design, there are two additional switches to add 1/3, 2/3, or full keyboard tracking of the filter frequency. When self-oscillating, the filter remains perfectly in tune with the other oscillators, and can provide a 'fourth oscillator' when required.
The filter output is fed into the Amplifier, which again has an ADS envelope, but with an added 'Release On' switch that allows the envelope to slowly die away at the same rate as its Decay setting once the note is released.
In the left-hand section of the panel are overall Tune and Glide controls, along with various knobs and switches for modulation. The 'Osc 3' button removes Oscillator 3 from MIDI key control to provide a fixed-frequency sound source, either for drones or, with its Mix slider turned down, for use as a modulation source for both filter frequency and the other oscillators -- as on the original, there are no dedicated LFOs. The Modulation Mix control lets you alter the balance of Oscillator 3 and noise, and you can control the overall amount externally using the modulation wheel found on most MIDI keyboards. With these few controls you can create vibrato, wah-wah and growls, as well as chiffy and breathy sounds.
And There's More
So far, apart from a few controls like Filter Type, the design is largely the same as the original. However, using the controls spread across the bottom panel strip (see page 104), each Model E you launch can run up to 64 voices spread across 16 MIDI channels and four stereo outputs. From left to right are a 16-LED display showing MIDI activity on each of the 16 available MIDI channels, and then a Channel selector switch to show the current front-panel settings for each MIDI channel. Next is an Output selector switch to route the currently displayed program to one of the four available stereo outputs; these are automatically added to the Channel Mixer when you activate a Model E synth, and can be seen in the main screenshot opposite.
The Program selector switch lets you choose any of the current 128 presets. You can also use MIDI Program Change commands, or click on the 7-segment display alongside that launches a large scrolling list of available presets. The current preset name is displayed on the panel, with Copy, Paste, and Compare buttons above it to help manage your sound banks.
Beyond these controls are two rotary knobs to add variable keyboard velocity sensitivity to the filter frequency and amplifier level, and three more knobs for the output. Panorama sets the position between left and right speakers, Spread adds a random pan offset to this, and Volume sets the overall level (which is also adjustable by MIDI volume). Of these three, only Spread is saved with each program. Finally, the Voices selector switch lets you choose the maximum number of voices available, up to a maximum of 64 (processor power permitting of course). You can override this at any time using the Mono switch on the main synth panel to revert to monophonic output.
Using Model E
Sounds can be loaded and saved in banks of 128 presets, and flicking through the five supplied sound banks demonstrated that Model E is certainly versatile. There are plenty of soaring lead sounds, solid basses, soft string pads, and percussive organs. A few to audition are the polyphonic glides of 'Moogstyle' (program 10 of Steinberg bank 1), the chord-on-a-note 'Autobahn' (program 36 of Steinberg bank 1), and the classic synth-brass of 'Mini Horn' (program 58 of the Hubertus Maas bank).
However, I found many of the supplied sounds disappointing. A fair few novelty effects are thrown in, and none of the banks contained the full 128 sounds (one of the two Steinberg banks only has 32!). However, I changed my mind about Model E once I started to program some of my own sounds: the 4-pole filter has a rich, natural and above all musical sound, with a lovely controlled resonance, while the 2-pole version sounds more electronic, making it more suitable for 'digital' sounds, but is definitely the poor relative. The envelopes can also be extremely fast when required (I measured an attack time of under 1mS), making Model E ideal for basses and plucked sounds. The addition of velocity control adds a lot to its expressive powers, and the overall sound is warm and smooth.
Overall, Model E is impressive, despite my reservations about its mouse-control system (see the 'Mouse Matters' box) and I expect it will sell in large quantities. It will appeal to those who want rich, traditional analogue sounds with the power of three oscillators. I don't expect a glut of Minimoogs to suddenly appear second-hand as a result, but if you want a taste of the real thing at a much cheaper price you should definitely give this a try.
Native Instruments Pro Five
Pro Five is, in essence, a software emulation of the Sequential Circuits Prophet 5 synthesizer. Like Model E, Pro Five is distributed worldwide by Steinberg, but this time is designed by Native Instruments, already well known for their Generator and Reaktor soft synths. They have used the same software engine to produce this VST Instrument, but since the elements of each voice are ha To gauge the amount of processor power required I played a static 8-note chord on both synths, and with my Pentium II 450MHz PC Model E consumed 18 percent and Pro Five 22 percent of available processor power. Both models managed 22 simultaneous notes before I started to get break-up due to processor overload, but switching to Model E's 2-pole filter mode let me increase this to 36 notes.
Both VST Instruments reviewed here have the same system requirements. Mac owners will need a Power Mac or compatible system with a minimum of a 604e/250MHz processor, MacOS 8.0 or higher, and Cubase 4.1 or other VST 2.0-compatible host software. In addition, Model E is the first soft synth optimised for the new G4 Velocity Engine. On the PC you will need a minimum of a Pentium 266MHz processor, Windows 95 or 98, and Cubase 3.71 or other VST 2.0-compatible host software. Both platforms will require at least 64Mb of RAM.
To gauge the amount of processor power required I played a static 8-note chord on both synths, and with my Pentium II 450MHz PC Model E consumed 18 percent and Pro Five 22 percent of available processor power. Both models managed 22 simultaneous notes before I started to get break-up due to processor overload, but switching to Model E's 2-pole filter mode let me increase this to 36 notes.
The original featured two oscillators per voice, five polyphonic voices and a five-octave keyboard. As with Model E, velocity sensitivity is added to the original spec, and its maximum polyphony of 32 voices is also much greater, although I suspect once again that few musicians will dedicate the amount of processor power required to achieve this. Unlike Model E, which simply demands that you occasionally insert your CD, Pro Five's protection system gives you the option of installing a 50Mb file called Mystery on your hard drive. If you do this you will occasionally be asked for the original CD-ROM to be inserted; if not you have to do this every time you start the program. NI are pleased that this system seems to stop Internet piracy, and most people won't grumble, but I hope I don't have to install too many more applications with huge protection files. The total install size is about 57Mb.
Each voice has two oscillators: Oscillator A has sawtooth and variable-width pulse (from 0 to 100 percent) waveforms, while Oscillator B has an additional triangle wave. Each waveform has its own on/off button, so you combine them in any way you please. Both oscillators can be tuned in semitone increments over a four-octave range using their Freq knobs. Oscillator B can also be detuned up to a semitone from A using the Detune knob, while a Lo Freq button lets you alter its frequency range down to 0.3Hz to 30Hz for use as an LFO. The outputs from the two oscillators, along with a white noise source, are mixed together using rotary controls in the Mixer. From here they pass into the Filter, which provides controls for Cutoff (frequency), Resonance, Env Amount, and variable Keyboard tracking. Finally the filtered signal enters the Amplifier. Both filter and amplifier have full ADSR envelope controls, and like Model E, the fastest attack times are under 1mS.
A comprehensive panel at the left-hand end of the panel contains modulation controls. There is a dedicated LFO with frequency variable between 0.04Hz and 20Hz, and a MIDI Sync button lets you quantise this to the nearest appropriate tempo via MIDI clock. The Poly Mod section lets you mix together the Filter Envelope and the output of Oscillator B, and use these to control Freq A, Pulse Width A, or the Filter frequency. This provides real versatility, since you can for instance thicken up the oscillator with envelope pulse-width modulation, or generate ring-modulator-like sounds by routing Oscillator B to the filter frequency. These possibilities multiply once you activate the Sync button for Oscillator A, allowing you to generate searing tonal sweeps as well.
The Wheel Mod section lets you send a mix of LFO and Noise to any combination of Freq A and B, Pulse Width A and B, and Filter frequency, with a depth controlled from the modulation wheel of your external keyboard. There are also various global controls scattered about the front panel such as Volume, Glide speed, a Release switch that toggles between the current ADSR release setting and minimum, and Vel to add a preset amount of MIDI velocity sensitivity to both filter and amplifier envelopes. MIDI activity is shown on a flashing 'LED' display, and you can also click on the Pro Five logo to launch a five-octave graphic keyboard, complete with Pitch and Mod wheels, to help when creating new sounds.
The number of voices can be set anywhere between 1 and 32, processor permitting, and a Unison button lets you create extremely fat sounds by using all voices to play a single note. It's wise to slowly increase the number of voices, as attempting unison mode with all 32 enabled is likely to crash the majority of computers: to give you an idea of the power required, my Pentium II 450MHz PC managed 27 voices in unison mode, with no other plug-ins running, before starting to glitch. Unlike Model E, Pro Five has a single stereo output, and you have to set up a pan position using the standard Cubase faders. However, if you want to go multitimbral you can simply launch more Pro Fives, each with their own stereo output.
To set overall tuning there is a Tune knob and an A440 tuning oscillator. An intriguing Tune button is also provided. In its 'on' state all voices are in 'digital mode' and perfectly locked in tune with each other. In the 'off' state voices revert to 'analogue mode' and each has tiny random variations added at various points in the signal chain to simulate the subtle pitch drifts found in most hardw
When I first launched Pro Five I thought Native Instruments had made a terrible mistake in providing such tiny knobs and buttons compared with those of Model E, but after using it for an hour of two actually found it far easier to use. This is because mouse-dragging lets you increase the current value using an upward movement, and decrease it using a downward one, making small changes comparatively easy. I much preferred this drag up/down mouse control to Steinberg's rotational approach, where the position of each rotary knob is altered by clicking on it and dragging the mouse in a circle to set the position of its pointer. While this lets you achieve very fine resolution by dragging in a larger circle, it is almost impossible to grab a knob without its current setting lurching wildly, which makes subtle editing extremely difficult. I found Pro Five in up/down mode a lot more controllable and playable than the Model E. However, if you prefer the Model E approach, holding down the shift key and clicking with the mouse on the NI logo switches the plug-in to this mode instead.
As on the original Prophet 5, programs are stored in eight Banks of eight, and up to eight Files of 64 presets can loaded at once, giving a total of 512 program numbers. A three-digit LED display is provided showing the current program number; alongside this are eight illuminated buttons numbered 1 to 8, along with two further ones labelled File and Bank, so that you can select any possible program number from 1 to 512. This 'octal' storage system can be confusing for those more used to the decimal approach, but accessing a particular program is quite quick once you get used to it. The full range of 512 programs can also be accessed using MIDI Bank Changes from 0 to 3, along with Program numbers of 1 to 128. You can load and save individual programs, Banks of eight programs, Files of 64 programs, or All 512 voices. You save your own edited programs by arming the Record button and then clicking on the destination program button.
Using Pro Five
Given the more comprehensive architecture, it's hardly surprising that Pro Five is more versatile than Model E, but the majority of sounds tend to have a rather different flavour as well. Pro Five's filter is harder and more metallic than that of Model E, which makes it perfect for clavinets, guitar, percussive organs, shimmering strings, and raspy brass sounds. The sync options let you explore the harder side of synthesis, while the versatile modulation options create some very unusual sounds such as echo, apeggios and ring modulation.
Only one bank of sounds is supplied, but it contains a full 512 programs, and very nice most of them are too -- even the majority of the sound effects are usable. Personal favourites were the analogue-sounding 'Low Dark Strings' (143), the echoes of 'Pad Oblique' (148), 'OldSynthyBrass' (152), 'Hammer Lead' (322), the spine-tingling resonances of 'PWM Sweep' (365), and the metallic twang of 'Muted Clav' (413).
If you want the Prophet sound inside Despite this, you still have to contend with the latency of your soundcard, which may impose a significant delay between pressing a MIDI note on your external keyboard and hearing the sound of your VST Instrument. Unlike stand-alone soft synths, VST Instruments cannot have their latency tweaked separately from the other audio tracks -- whatever latency value is reported in the Cubase VST Audio System Setup is what you get for everything. A value of 48mS will feel extremely sluggish, and 24mS is about the highest latency that is usable, particularly when playing sounds with a fast attack. Those with latency values of 10mS or less will find the real-time response as good as most stand-alone MIDI keyboards or modules. The upshot of this is that having well-written ASIO drivers for your soundcard is an absolute necessity. Even PC owners like me who are well pleased with the real-time performance of their PC soundcards with a stand-alone softsynth such as Reaktor or VAZ Modular may still be frustrated when running VST Instruments. For instance, my Yamaha SW1000XG performs well in a PC with both these applications when you choose its DirectSound drivers, but in Cubase VST its ASIO drivers can only manage 106mS latency. Bear these limitations in mind before getting out your credit card.
Timing & Latency
VST Instruments offer far more accurate timing than hardware MIDI instruments, since event positioning is accurate to a single sample. At 44.1kHz this equates to 23 microseconds, 42 times better than the 960 microsecond gasp between two MIDI notes.
Despite this, you still have to contend with the latency of your soundcard, which may impose a significant delay between pressing a MIDI note on your external keyboard and hearing the sound of your VST Instrument. Unlike stand-alone soft synths, VST Instruments cannot have their latency tweaked separately from the other audio tracks -- whatever latency value is reported in the Cubase VST Audio System Setup is what you get for everything. A value of 48mS will feel extremely sluggish, and 24mS is about the highest latency that is usable, particularly when playing sounds with a fast attack. Those with latency values of 10mS or less will find the real-time response as good as most stand-alone MIDI keyboards or modules.
The upshot of this is that having well-written ASIO drivers for your soundcard is an absolute necessity. Even PC owners like me who are well pleased with the real-time performance of their PC soundcards with a stand-alone softsynth such as Reaktor or VAZ Modular may still be frustrated when running VST Instruments. For instance, my Yamaha SW1000XG performs well in a PC with both these applications when you choose its DirectSound drivers, but in Cubase VST its ASIO drivers can only manage 106mS latency. Bear these limitations in mind before getting out your credit card.
Just Like The Real Thing?
I've played both Minimoogs and Prophets in my time, so I was interested to see how close these software simulations came to the sound of the originals. Each hardware revision of both Minimoogs and Prophets sounded slightly different, so there is room for variation, but there is no doubt in my mind that these two software synths have captured the essence of the instruments that inspired them.
Like the originals, the VST Instruments each have a unique character, and however hard you try, it's impossible to duplicate the sound of one with the other. Apart from the differences in architecture, this is largely down to the sounds of each filter. Model E's virtual 'ladder' filter design produces subtle distortions that, like much valve circuitry, add a warmth to the sounds passing through it; for this reason it's disappointing that other audio tracks can't be routed through it, even as a separate audio filter plug-in. On the other hand, Pro Five has a more electronic sound, but is a classic in its own way.
Mind you, there is more to creating a realistic analogue sound than the response of the filter. One advantage of having the real instruments in front of you is that you can adjust their controls in real time to suit the song and your mood. Changing the filter frequency, resonance, or envelope depth on the fly can be as much a part of an analogue synth performance as using pitch-bend and modulation. In most cases these VST Instruments' 'analogue' experience is reasonably believable, but although you can automate every Model E and Pro Five parameter using MIDI controllers, this inevitably means using 128 stepped values -- you cannot sweep them as smoothly as their analogue equivalents. Thus whereas sweeping the filter frequency with the envelopes is very smooth, doing the same thing using MIDI controllers restricts the number of steps to 127, making the filter frequency jump in semitones. On the other hand, the ability to automate each and every control via MIDI does provide huge performance possibilities.
Like many soft synths, both Model E and Pro Five also suffer from subtle aliasing distortion when playing notes higher than three octaves above Middle C (ie. with a fundamental of
MODEL E £149
Rich filter with resonance.
Up to 16 multitimbral synths and 64 voices from each Model E.
Four stereo outputs.
Infuriating mouse control.
No external filter input.
Steinberg's Model E may not look exactly the same as its inspiration, and its sound doesn't quite match the original, but both are close enough to make it a highly desirable purchase.
One area where a software-modelled synth can improve on the original analogue version is that the developers can easily add extra features. There are those who insist that any recreation of an old analogue synth should adhere closely to the original design. However, I found the much increased polyphony of both soft synths truly liberating, while the velocity sensitivity of both amplifier and filter gave far more expressive possibilities (and of course you can turn it off if you want). Pro Five seemed complete in itself, but to me Model E cries out for the addition of a dedicated LFO, an oscillator sync option, and PWM, to give it a similar range of sounds.
Having VST Instruments at your disposal within a MIDI + Audio sequencer is, as you might expect, more convenient than running a separate application, and although many stand-alone soft synths provide built-in effects, there is nothing quite like having your complete arsenal of VST and DirectX effects available to patch in at will. The 'hard-wired' approach of these two designs also lowers processor overhead compared with most modular ones, and you can always see the total overhead taken by synths and sequencer on the VST Performance meter.
Of the two synths here, Pro Five is undoubtedly more versatile, but does not offer three-oscillator sounds and the famed Moog filter sound. If you are going to buy one or the other, you should be guided by the sort of sounds you want. There is little to choose between them in other areas: although you can get 16-channel multitimbral music across four stereo outputs from a single Model E, you can run multiple Pro Fives in parallel if required to achieve the same end result. There's no denying that they are both versatile and easy-to-use synths at a very reasonable price.