Virtual Instrument RoundUp

From recording tips to gear reviews, here’s what you need to know about using musical instruments that exist only in the virtual world.

FROM RECORDING TIPS TO GEAR REVIEWS, HERE'S WHAT you need to know about using musical instruments that exist only in the virtual world.

WTF happened?!? Instruments, particularly keyboards, used to be things that you set up in your studio and played into a recorder. But now, they’re in your recorder—and instead of arriving in a wood, metal, or plastic enclosure, they come on a CD- or DVD-ROM, or are downloaded from the Internet. Really, these are instruments?

Really, they are. There are over a thousand virtual instruments available, ranging from questionable to insanely great. We already did a roundup on virtual drummers in the 07/07 issue, but now it’s time to lay our fingers on some synths and samplers.

Of course, lines of code and a cool user interface do not an instrument make. One of the most important addons is a control surface to give physical control over an instrument, whether a dedicated box like Native Instruments’ Kore, or a more general purpose “fader box” controller. Either of these help restore the physical element to virtual instruments.

The computer comes into play, too. Today’s fast computers reduce latency, and make the playing experience far more enjoyable; also, companion editing applications reveal an instrument’s innards in a way that’s hard to pull off in the physical world. Truly, the virtual instrument has come of age.

This roundup has two main sections. The first covers tips and techniques on recording virtual instruments, because the process is not always as obvious as it might seem. The second features reviews of several current virtual instruments, and frankly, this presented a problem: There are so many of them we couldn’t cover even 1% of what’s out there. So, we chose a selection that’s fairly representative of what you’ll find— from clever analog synth emulations, to super-synths with huge sound libraries.

There’s a third part, too, but this one threw us a curve: We wanted to get some insights from prominent virtual instrument designers on the state of the art, so we asked Ernst Nathorst-Böös (Propellerhead Software), Doug Rogers (EastWest), Peter Gorges (Advanced Instrument Research, a division of Digidesign), and to add a non-designer viewpoint, Nick Batzdorf (Editor of Virtual Instruments magazine) to give us their thoughts. We expected to get a few useful quotes we could use in sidebars, but to our surprise and delight, we instead received long, detailed, and introspective responses from all of them. There was more than we could fit in the magazine, and editing them seemed just plain wrong. So, we’ve put the complete, unedited versions of these interviews on the web at . It’s rare to get these kind of insights from people who are, in various ways, responsible for the virtual revolution we’re experiencing in music.

It was a blast putting this roundup together, and we sincerely hope you enjoy it. Play on!


Well, it’s easy, isn’t it? You just insert it into your host and click on record. Right?

Wrong, because a virtual instrument should be an instrument, not just another feature in your DAW. It should feel right, and offer expressive control options.


Most people understand what “latency” is with virtual instruments: the time it takes between hitting a note and hearing it. The greater the latency, the less satisfying the playing experience.

Aside from the obvious solution—a wicked fast computer, and well-written audio drivers for your audio interface— there’s another approach I call “variable latency recording.” The premise is that latency is a continuum between low and high latency, and the trick is to find the “sweet spot” along that continuum for the task at hand. For example, when recording, you want the lowest latency for the best feel, but when mixing, you can use higher latencies to accommodate all the CPU-sucking tracks, instruments, and effects.

When recording a virtual instrument, you need to unload your CPU of as many obligations as possible so it can devote all its power to the instrument. Here are some easy ways to deal with latency issues.

• Record keyboard parts early in the recording process, when there are few other virtual instruments or effects. You’ll be able to get away with low latency.

• For “guide” tracks, use audio tracks (recorded instruments or loops), as they require less CPU power than virtual instruments.

• Record with any “low-resolution” options enabled (Figure 1). These days, fewer instruments have low-res modes because computers have gotten so fast—but those that do often allow for lower latency when recording in low-res mode.

• Don’t use CPU-hungry effects, like convolution reverb. Add these during mixdown, when it’s possible to increase latency to 20ms or even more and not have the delay be a problem.

• A “workstation” (or multitimbral sampler) plug-in loaded with instruments will use less CPU than an equivalent number of individual instruments.

• The Freeze function is your friend. Most modern DAWs can freeze a track, which renders the instrument audio to a temporary (or permanent, if you want) hard disk audio track, then “disconnects” the virtual instrument itself from the CPU to save power. This is an excellent option when you’re partway through recording; you can freeze your virtual instruments, yet still set a fairly low latency when recording overdubs.

• Don’t have other programs running, even in the background, when recording. They all demand at least a little bit of RAM, CPU power, or both.

• Consider the new generation of hardware synths that think they’re software. The Korg M3, Yamaha Motif XS, Roland Fantom G, and several others have software GUIs that run as plug-ins in your computer (Figure 2), and communicate with the hardware via USB. For all practical purposes, the synth is a plug-in—except with a bitchin’ control surface, and on the downside, you can’t instantiate more than one instance. But as these are multitimbral anyway, this isn’t much of a limitation. These synths typically include onboard mixing and effects, which takes even more of a load off the CPU. One of the more interesting examples of this approach is Cakewalk’s Sonar V-Studio, whic builds a hardware Roland Fantom synth into the audio interface (Figure 3). Thus, you have a multitimbral hardware synth available at all times during the recording process that doesn’t require any significant amount of CPU.

There are other hardware solutions, too. The Use Audio Plugiator (see the associated review on page 52) basically houses several synths in DSP, and communicates with its GUI via USB; the venerable SCOPE system (formerly from Creamware and now from Sonic Core) has several DSP-based instruments as part of their “virtualized studio” system, and TC Electronic’s PowerCore has a synthesizer option.


With so many instruments these days having huge sound libraries, there’s the question of where to store all that data—not a trivial issue, when you consider that having only Spectrasonics’ Omnisphere and MOTU’s Electric Keys requires over 80GB. (And yes, they both take a long time to install—but six double-layer DVDs is way better than 27,000+ floppy disks!)

Most installations default to installing the sounds on your root drive, usually as some sub-folder of the program folder. Not only does this “crowd” the all-important OS drive, but streaming samples from disk is a diskintensive activity.

Therefore, it’s a good plan to have a separate, huge drive dedicated solely to instrument libraries. I installed a 1TeraByte SATA drive inside my computer, but another option is to use a fast external FireWire or USB 2.0 drive. In general FireWire provides somewhat greater sustained throughput than USB 2.0, but has a reputation of being more “finicky” in terms of getting along with particular chip sets.

If you place the sounds somewhere other than installation default, you’ll probably have to tell the program where to look for its library. There may be the option of searching for files, but if you have a big hard drive, that can take a while—it’s faster to point to the sounds manually.

Different instruments handle this in different ways. The most common is to specify a file path, typically in a Preferences menu (Figure 4). Another method is to create aliases where the sounds are assumed to be, then point to where the sounds really are (Figure 5). However, the shortcut must have the exact name—you can’t leave “shortcut to” or “alias” in the name.


Although you can always add controller data after recording, it’s more spontaneous to do it while playing. Many keyboard controllers include some sort of control surface, as do hardware synthesizers (Figure 6). Lacking that, generally you can use a fader box or other MIDI control surface designed for mix control, like the Novation Nocturne, Mackie Control, Native Instruments Kore (or their Maschine controller for MPC padbased instruments), Behringer BCF2000, and the like.

The control surface generates MIDI data. If there’s a keyboard, you’ll have velocity, mod wheel, and probably aftertouch as modulation sources. There may be other options, like an expression pedal input, or front panel controls.

As to actually linking physical controls to virtual instrument parameters, the most common option is to simply right-click or ctrl-click on a parameter to bring up a “learn” menu: Move your hardware control, and the instrument will link that control’s movement to the selected parameter (Figure 7). You will likely select one or more of the following options for optimizing the response.

• Polarity or Invert. When positive, increasing the control’s value increases the parameter’s value. When negative, increasing the control’s value decreases the parameter’s value. As a practical example, suppose you want to increase “Drive” (i.e., add some distortion to make the sound dirtier) with a mod wheel—except that this creates a volume jump. By applying negative polarity to the volume over a limited range, you can turn down the level simultaneously while increasing the drive.

• Range, Depth, or Level. This determines how much modulation will be sent to a parameter, thus determining the range of the parameter the hardware control will sweep. You might want the control to cover a parameter’s full range, or only a small portion (e.g., filter resonance from 5% to 30%). Often polarity and range are combined in a single control (Figure 8), where the center position indicates no modulation, turning clockwise gives positive modulation, and turning counterclockwise gives negative modulation.

• Smoothing. Because incoming MIDI data is usually quantized into 128 steps, this may produce an annoying “stair-stepping” effect instead of the smooth, continuous control you’d expect from an analog synthesizer. Smoothing integrates these values internally, within the synthesizer, to smooth the response. The only drawback is that this might add a bit of an attack time, and a slight “overhang” during the decay.

• Response curve. The parameter may respond in a linear, log, or exponential fashion.

Even if you don’t have a physical controller, all is not lost: With many instruments and hosts, you can enable automation and write automation data by turning a virtual instrument’s control with a mouse. Of course you’re limited to controlling only one parameter at a time this way, but it’s better than nothing, and you can always overdub automation passes.


With so many virtual instruments, it was very tough choosing just these few—but we feel each one represents a particular genre of instrument. Most have downloadable demos or at least online audio demos, so with few exceptions, you can “try before you buy.” As with our other roundups, all products are presented in the order our layout artist thought looked best on the page, and all prices are list prices.


Image placeholder title

Electric Keys is a sample playback-based VST/AU/RTAS/DXi/standalone encyclopedia of 50 vintage keyboards—combo organs, electric pianos, clavs, organs, piano bass, string synthesizers, “tape samplers”—with more presets than I feel like counting. It uses the UVI audio engine featured in other MOTU instruments, with an efficient browser and several sound modification options.

Included sample library: 39.1GB

Copy protection: iLok.

The interface: A basic “shell” has controls for Volume, Tune, Bass, Middle, Treble, Drive, Tremolo (with off/on/pan switch, Depth, Speed, and Sync), and four velocity curves. Loading a different instrument changes the skin to reflect the look of that particular instrument.

You can load up to two sounds simultaneously, each with its own volume, pan, and mute control; an additional “edit” panel for each instrument provides significant editing, including 14 filter options with envelope, amplitude envelope, and pan modulation. Also useful: The “effects rack,” with an amp simulator (that goes beyond just amp sims), filter, phaser, flanger, chorus, “vinyl,” delay, and reverb. There’s also a standalone mode.

Strengths: It really does have just about everything you could want from a set of vintage keyboards, including sampling with fine fidelity. Although the editability isn’t spectacular, neither was the editability of the keyboards being emulated. The best part here is the filter—14 different filter types—although the effects can also add a very useful dimension, and even be saved as presets that are independent of instruments.

Limitations: There’s limited velocity multisampling on several of the keyboards; for example, electric pianos with five different velocity samples per note. Also, for authenticity keyboard ranges are limited to that of the original keyboard— but I would prefer if the ranges were stretched a bit further. Sometimes playing an instrument out of its natural range gives cool timbres.

The bottom line: If you lack vintage keyboard sounds, Electric Keys will plug those holes—it reminds me conceptually of Big Fish Audio’s Drums Overkill!, which sampled every drum machine created since the Bronze age. Even if you have most of the sounds represented here in other sample libraries, there are enough variations that you’ll surely find ones you don’t have. Overall, the completeness and relatively low cost are in Electric Keys’ favor.

Price: $295


Image placeholder title

Combining both hardware and software, Maschine is a package that streamlines the beat creation/groove process. It follows the usual drum machine paradigm (you string patterns into songs), but also borrows Ableton Live’s “scene” concept for triggering multiple patterns simultaneously. It’s also a sampler and re-sampler that can slice files.

Included sample library: 4.75GB

Copy protection: Online registration.

The interface: The hardware is a classy, compact playing surface with 16 extremely predictable pads, 41 buttons, 11 rotary encoders, and two LCDs; while optimized for Maschine, you can also switch it into a MIDI controller mode.

The software includes two applications. Maschine Editor manages sounds, patterns, and scenes—assign sounds to pads, apply effects to individual sounds or groups, sample, create rhythm patterns, and combine patterns into scenes. Controller Editor allows programming the controller’s pad and button MIDI assignments when used as a MIDI controller. Most functions are duplicated in hardware and software, so you can adjust a parameter either physically or virtually; the manual often describes how to do a function using either option.

Strengths: Maschine is much deeper than it appears, although you can get a lot out of just the top layer, and the sample library is packed—fortunately, the browser has categories and tags. The software/hardware integration is as good as it gets, and the effects are huge: While there are the expected effects (EQ, compressor), most seem designed specifically for percussion (e.g., lo-fi, unusual reverbs, frequency shifter, etc.). You won’t find them in your typical host program, so it’s good Maschine has them. You can automate most of their parameters.

Limitations: This isn’t a limitation of Maschine, but what would be acceptable latency with other instruments can become unbearable with percussion. It won’t matter if you’re into step sequencing, but for realtime playing, a fast computer is essential. Also, Maschine won’t do MIDI out so you can’t send the patterns to anything else, nor can the notes that trigger slices be brought into a host; according to NI, these features will be in a future update. There’s also no REX file support.

The bottom line: No single element of Maschine is particularly innovative, but put all the pieces together, and you have a groove/beat machine that is truly fun to use. If beats are what you do, this machine was designed with you—and tactile, realtime control—in mind.

Price: $669


Image placeholder title

Playa is sort of a hip-hop construction kit plug-in (that also works in standalone mode) with a boatload of samples, MIDI compatibility, and an MPC-style pad interface—which, incidentally, works very nicely with Maschine’s controller. Editing is limited, but Playa is quick, fast, doesn’t draw much CPU, and has a high “fun factor.”

Included sample library: 5.5GB

Copy protection: Online authorization via virtual PACE.

The interface: Playa’s graphics are all about being a DJ/groove-style instrument, so instead of having little teeny controllers with illegible type, you get obvious faders and knobs that look more like a DJ mixer. Almost all of these respond to MIDI control and/or automation, so with a decent control surface, you have more of a “live performance instrument” than a plug-in.

In addition to being able to load particular instrument sounds (drums, guitar, bass, keys, choirs, synths, orchestral, and more—all the usual suspects), you can also load particular pad layouts. Note that Playa isn’t multitimbral, so it’s one sound/one output pair per instance. But you can load up on the instances—Playa doesn’t need a lot of CPU juice.

Strengths: Solid, mostly genre-specific samples form a solid foundation. Playa is very easy to learn, once you figure out a few basics (like how to choose keyboard or pad input, and the difference between instruments and layouts). And, the automation/MIDI control allows for expressiveness.

Limitations: The only effects are chorus and delay, so figure on adding plug-ins afterward on the host. Also, Playa doesn’t really work that well for standalone operation, because it loads only one sound at a time—you can’t, for example, lay down different tracks and overdub against a pattern. And of course, it’s designed specifically for groove/hip-hop type sounds, although I feel many sounds are highly suitable for dance music.

The bottom line: Playa costs less than the average virtual instrument, with the tradeoff being less editing and complexity. You still get a fine sample library, simple operation, and that MPC-style vibe. Some consider it “Maschine lite,” but they’re very different animals: Playa is about sounds, not sequencing, slicing, or using a custom controller.

Price: $149.95


Image placeholder title

Plugiator comes with four virtual, DSP-driven synths in a compact, tabletop hardware box. You can take it on the road and use it live, or feed it into your DAW of choice. Furthermore, its Plug-In Manager software editor takes advantage of Plugiator’s USB interface for preset management and the ability to tweak the instrument parameters beyond the five parameters brought out to realtime control knobs on the unit itself.

You can’t instantiate Plugiator as a plug-in within a DAW yet (Use Audio is about to introduce VST front ends for the various Plugiator instruments so that they can be integrated with a DAW as plug-ins), but you can send MIDI data from your DAW to Plugiator via USB, and feed Plugiator’s audio outputs into your audio interface’s inputs.

Included sample library: There’s no library per se; Plugiator ships with four modeled instruments (Minimax virtual analog synth, LightWave wavetable synthesizer; B4000 Hammond B3 model, and Vocodizer vocoder). Other instruments are available for $49 each: Prodyssey, Pro- 12, FMagia, and Drums’n’Bass (drum machine and bassline).

Copy protection: Online registration.

The interface: The hardware has 1/4" left and right outs, a 1/4" stereo headphone jack, balanced mic input for the Vocodizer, and USB. However, you can’t plug in a USB MIDI controller; you need to use the 5-pin DIN MIDI connectors, or plug a USB keyboard into your computer, then send its data via USB to the Plugiator. The Plug-In Manager application is fairly standard, throwing all the various instrument parameters on-screen.

Strengths: It sounds fine, doesn’t load down your CPU, and is platform-independent: If your DAW can spit out MIDI, Plugiator knows what to do with it. The realtime tweaking controls are a nice touch, and extra instruments are inexpensive—for $698, you get eight instruments in a small, portable package.

Limitations: Although you can load eight instruments in DSP, you can only use one at a time. Also, each instrument is limited to 100 patches. The sole effects are delay and chorus, and different instruments provide different amounts of polyphony (e.g., ten voices for the Minimax, 91 for the B4000). Finally, you can’t run the Plug-In Manager while running a DAW unless you use a MIDI loopback applet, like MIDI Yoke.

The bottom line: Plugiator takes an unusual approach to virtual instruments, but it’s an effective one. There’s also a certain feeling of permanence; if Apple or Microsoft updates their OS you’ll still be able to use it, and it makes a reliable live performance device.

Price: $549


Image placeholder title

It takes a lot to add something new to analog emulation synths, but Future Audio Workshop seems determined to do that with Circle. While the sound-shaping elements aren’t out of the ordinary—oscillators with analog waveforms or digital wavetables, various effects, filtering, and multiple modulation sources (LFO, envelope, and step sequencer)— the way they’re put together is clever. I also like the tuned delay that feeds the output back to the input through a “Feedback” section, which is sort of like an extra voice.

Included sample library: There are no samples, but the preset management includes tagging/categorization and search. FAW has posted additional free sound sets on their website.

Copy protection: Online authorization.

The interface: First of all, the interface looks modern rather than emulating a vintage vibe—it reminds me a bit of Live crossed with Tracktion—and features intelligent use of color. For example, you can drag a colored circle from a modulation source to a controllable parameter (e.g., filter frequency), and the parameter’s “virtual jack” acquires the same color. Each modulation source is color-coded, making it easy to see what connects to what without using a modulation matrix. There’s also a little ring around the “jack” for adjusting modulation depth, like Arturia’s Moog Modular.

Strengths: Circle is exceptionally easy to figure out and use. I looked at the manual only to check whether I’d missed anything (I hadn’t), but I must say they did a really good job on the documentation. If fact, if I ran an electronic music school, I’d buy a site license and use Circle to teach synthesis.

Limitations: You can’t load your own wavetables, and all the envelopes are basic ADSR types—okay if you’re going for straight analog emulation, but they’re pretty old school compared to things like tempo-synced envelopes. The LFOs offer 16 different waveforms, which is appreciated—but again, you can’t load your own—and the step sequencers aren’t as complete as, say, Cakewalk Rapture.

The bottom line: If you like the idea of analog emulation but don’t want something that’s just an imitation Minimoog, it’s worth downloading the Circle demo to find out what the fuss is about. It’s easy on both the eyes and ears.

Price: $199


Image placeholder title

Do you like a DX7 Rhodes better than a real one? Do you wish your guitar string had editable parameters? Do you think Lounge Lizard really nailed the electric piano sound? Then you’re a candidate for String Studio VS-1, a VST/AU/RTAS/DXi plug-in (standalone too) that’s based on physical modeling of strings rather than samples. Not only can you create recognizable string sounds like guitar, bass, and clav, you can also make sounds that exist only in your imagination, like stratsichords or clavbasses. (By the way, if you’re a Live fan you’re not having déjà vu: String Studio is very similar to Tension, an optional Ableton instrument for Live 7.)

Included sample library: No samples, so there’s no sample library—but you do get lots of presets. They’re useful in their own right, and also provide a great point of departure for your own tweaking.

Copy protection: Online challenge and response.

The interface: It looks like something out of Captain Nemo’s submarine, but that’s a compliment—the look is retro, but in a Moog sythesizer sort of way. You can choose different excitation types (pick, bow, and hammer), damping, string parameters, fret/finger interaction, and soundboard bodies, and you also have synth elements like filter, filter envelope, LFP, EQ, and distortion. Everything is brought out to switches, controls, and drop-down menus that respond to MIDI control and automation.

Strengths: Modeling-based synthesis allows more expressive possibilities than sample-based synthesis, because samples can be altered only so far without sounding freakish. And unless you own the Scope system Six-String plug-in, there won’t be anything quite like this on your hard drive—you’ll be exploring new sounds every time you twist a knob.

Limitations: There’s not much to dislike: String Studio is cost-effective, original, delivers useful sounds, and allows for lots of creativity. You can complain that it’s a one-trick pony, but it’s a really cool trick—and the product lives up to the company’s promises.

The bottom line: If you’re looking for a workstation-type of plug-in, this is about as far from that as you’ll get. However, if you have good line-up of plug-ins and are itching for something different, you’ll likely appreciate the original, musically-useful approach String Studio brings to the table. Sure, you can just call up presets; but this is an instrument that’s worth tweaking to get the exact type of sound you want.

Price: $229


Image placeholder title

This is based on the multitimbral SampleTank engine, with 16 channels and 16 stereo output pairs; you can also save and load preset combis. In addition to 17 instruments (Mellotron, Chamberlin, Optigan, Novatron, Stylophone, etc.), there’s a wealth of effects—32 total—that add serious value and sound-shaping ability. You’ll find considerable editability, from filtering to stretching; you can apply these edits selectively to samples in different keyboard zones.

Included sample library: 2.4GB of meticulously-produced samples.

Copy protection: Online authorization.

The interface: It’s basically SampleTank with a different skin that has the beat-up vibe of a vintage instrument. The upper right half has a browser; a multitimbral map to the left can switch between channels 1–8 and 9–16. The lower half has multiple pages for accessing and editing parameters, with synth engine options and effects parameters. A virtual keyboard lets you play notes.

Strengths: SampleTron can emulate vintage instruments exactly, including quirks (the 8-second time limitation of Mellotron “loops,” tape hiss, etc.) or you can dial back the lo-fi and also do infinite looping, like a modern sampler. The effects are very useful, and overall, SampleTron is far greater than the sum of its parts. The correct judgment calls have been made on how to bring these instruments into today’s world, while being respectful of their rightful place in musical history.

Limitations: VST, AU, and RTAS automation is limited to controlling only six parameters—volume, pan, and the four macro controls. This is mitigated by being able to control almost all parameters via MIDI controllers, so if you have any kind of MIDI control surface, you’ll be okay.

The bottom line: I’m not a huge fan of those old tape samplers, because the sound usually left so much to be desired. Also, they stereotyped your music—“Moody Blues strings,” “Strawberry Fields flutes,” etc. But while this package can do lo-fi if desired, Sonic Reality has done the audio equivalent of restoring an oil painting: The dust and dirt is gone, but the character remains, and you see details you’ve seen before. In the process, they’ve managed to make these instruments relevant today—it’s not just about nostalgia. That’s a noteworthy accomplishment.

I’ve never used any samples of this type of instrument in any of my music before. Much to my surprise, SampleTron will change that.

Price: $229.99


Image placeholder title

Spectrasonics has never stumbled, while producing exceptional virtual instruments. So perhaps it’s not surprising that it would take Spectrasonics to outdo Spectrasonics, and this flagship instrument delivers on all levels.

Sure, Omnisphere deserves a long and comprehensive review; I recommend the one in the 12/08 issue of Keyboard, because I basically agree with everything it says. But there’s one caution: If you’re expecting a super-GM module with traditional sounds, Omnisphere is not the droid you’re looking for. While there are some standard instruments, Omnisphere is about pushing the envelope, not copying it.

Included sample library: 42GB and over 6,000 patches.

Copy protection: Online authorization.

The interface: There are four primary pages for editing. Main handles the crucial controls for the two layers per patch you can load simultaneously, Edit is for deep editing, FX calls up 12 effects (four for each layer, and four “common” effects), and Arp is a 32-step arpeggiator. That’s for individual patches. Go into Multi mode, and Stylus RMX fans will feel right at home: eight channels for eight patches, each with level, pan, and four aux sends for feeding up to four effects per aux, as well as four “master” effects. But you’ll also find a “live” mode for on-the-fly stack editing and playing, and a Stack mode where you can set up instrument crossfades, velocity switching, splits, and the like. All of this is obvious and straightforward— no small feat with an instrument on this level.

Strengths: While many other synths just add features (and Omnisphere has plenty of those), it’s the sound library that makes Omnisphere unique. Yes, there are some guitars, strings, and synths, but if you get a feeling of “been there, done that” when playing with soft synths, you won’t feel that way with Omnisphere. Many sounds are sufficiently inspiring you’ll be writing a song before you know it. Omnisphere’s sample library oozes creativity, and a browser can help narrow your searches.

Limitations: There’s no standalone mode, so live, you have to use a host. You can’t load your own samples (but the company isn’t ruling out that possibility in a future release), and there’s no demo version.

The bottom line: Reviewing this in 2,500 characters is like reviewing Paris by saying “It has this really tall tower left over from a world’s fair, lots of museums, and great restaurants.” Paradoxically, though, despite its sophistication Omnisphere is easy to summarize: It’s a tour de force of sampling-meets-synthesis that’s surprisingly easy to use, extremely playable, cost-effective, and inspiring.

Price: $499 (upgrades available for Spectrasonics Atmosphere users)


When I wrote the virtual instruments roundup for EQ magazine, I sent questions to four industry luminaries so I could have a few quotes to sprinkle around the roundup, sort of like tips. Little did I know that each would provide me with in-depth, insightful comments that really deserved an article of its own.

Rather than try to extract just “the good parts” of the article—an impossible task, given that there were only “good parts”—all the interviews are being published here in their entirety. The four participants are Doug Rogers, from EastWest; Ernst Nathorst-Boos of Propellerhead Software; Peter Gorges, formerly with AIR (although he was still with them at the time of this interview) and founder of Wizoo; and Nick Batzdorf, Editor/Publisher of Virtual Instruments magazine. I hope you enjoy their comments as much as I did.


CA: Why did you feel the need to develop your own instrument playback engine? Was it a question of technology, licensing, both, or…?

DR: Well, we got the idea to develop our PLAY software after the software licensing situation that we had with Native Instruments wasn’t working out. With Native Instruments we were using generic software that we couldn’t customize, and because we have many products that demand some form of customization to really make them work as we intended, that wasn’t a situation we could live with for long. So we made the very hard and expensive decision to start our own software division so we, not some other company, could control our future.

CA: You were one of the first companies, if not the first, to partner with a virtual instrument company (Native Instruments) and wrap a playback engine around sound libraries. What caused you to move in this direction—fewer compatibility issues, more difficult to steal, more flexibility, or all of the above? What was the seminal moment when you thought, “I know! Let’s make our libraries into virtual instruments!”

DR: I think all of the above is applicable. Our Symphonic Orchestra was the first of our products to be bundled with a software sampler, and it was so popular we never considered going back to creating the multiple formats of the same sample library for hardware samplers. Thank goodness those days are over!

CA: You were also one of the first companies to take 64-bit operating systems seriously. What do you see as the main advantage of 64-bit operation? Clearly, the rate of adoption of 64-bit systems has been slow, with a few companies like Cakewalk being the exception. Do you think adoption will accelerate, or just build slowly over time?

DR: In the beginning, because the technology was in its infancy there were serious limitations. You were mainly dealing with importing raw samples into hardware samplers such as Akai, E-mu, Roland etc., and you couldn’t do that much with them due to a lack of memory. I remember our first piano collection that got a five-star review had two dynamics, and you needed two fully loaded Akai samplers to play them back! That all changed with the introduction of streaming data from personal computer hard drives; we were involved in the development of that technology. Now you can do a lot because of powerful personal computers and 64-bit technology, which as you say is in its infancy. We introduced 64-bit Virtual Instruments to the market two years ago. At that time only Cakewalk Sonar supported 64-bit. Now Steinberg/Yamaha Cubase supports 64-bit, and we always offered a 64-bit standalone version and 32-bit compatible version as well—that is necessary when you are pioneering new formats.

The advantages for the user will force the remaining companies to get with the program or lose market share, and we fully anticipate 32-bit applications will disappear over the next few years. 64-bit systems make 32-bit systems look like dinosaurs. While most high-end computers will support 16GB, a very small set of high-performance enterprise computers (with Microsoft Vista Ultimate) can already use 128GB of RAM. As hardware develops, the theoretical limit is 16 Exabytes, or 17 billion GB! However, with that kind of complexity also comes the difficulty of trying to make it all work on such a multitude of systems such as the thousands of combinations of computer hardware and software that users have. It’s just not reasonable to expect any software to work on all of these systems (some of them built by the users themselves), so we concentrate on making the software work on the most popular combinations and expand that over time to less popular systems.

To lower the risk for our customers, we have “fully functioning” trial versions of some of our products so they can see how well they perform on their computers prior to purchase. Fortunately after three years of development PLAY is mature, and performing well on most systems, and in 64-bit mode, outperforming the competition. The benefit of moving from these hardware units to computers is, our customers now have these huge detailed instruments with up to 16 dynamic layers, and many articulations controlled by keyswitches and other controllers, and can do things with them that we literally could never have dreamed of 15-20 years ago. With our built in scripting engine in PLAY (soon to be available to our users in our PLAY PRO programming editor), we literally can interpret their playing style. It’s heady stuff!

CA: Your Fab Four library is impressive for many reasons. But I can’t help wonder how long it took to put it together, and whether you expected to make money off of it or whether it was more a labor of love? Could you describe the process of creating this library? Go into as much detail as you want!

DR: When I started the Fab Four project I didn’t really care if I made a profit off it or not because it was just something I wanted to do, and I figured that after putting 20 years of my life into this industry that I could do something for me for once. The interesting thing is that it actually broke even the day it was released, and it’s been a huge financial success to the point where I’m thinking of doing a follow up. Sometimes when you approach things without the profit motivation they end up being some of your most successful work. So initially it was just something I wanted to do for me, and I thought that there would be interest in it, but not as much as there was. It’s been a really successful product. Of course it also garnished the MIPA award—which is sort of like a Grammy in our industry because it’s voted on by over 100 international music magazines. So to win that means there was a consensus by others in the industry that we pulled off what we attempted to do. It started off as a hobby and ended up being great for our business. I think that’s how a lot of things develop. If you love what you do, then typically that can be passed on to other people for their enjoyment as well.

With regards to the Fab Four sounds, I am a Beatles nut, they were the initial influence for my fascination with sounds, however the Beatles obviously didn’t invent those instruments: Rickenbacker, Ludwig, Epiphone, Gretsch, Vox, etc.; commercial manufacturers made them. It was the sounds they and their engineers came up with that became iconic. It was really obvious to me that if I didn’t have exactly the same kind of equipment that they used, including all of the recording equipment, that there was no way I was going to be able to get that sound. And that’s what took a lot of the time in the pre-production stage of planning Fab Four. I probably put over a year of work into researching, finding, and procuring the equipment that we needed to do it.

The people I worked with on the project also had a strong connection with the Beatles. Ken Scott worked on five Beatles albums and was first engineer on two of them. The drummer and guitarist worked with Paul McCartney. So they were taught how to get the sounds and how to play them live. What is interesting is, the music that has been sent to me from users sounds nothing like the Beatles; our users are blending these classic sounds with other styles to give them flavor, and it works. The demos we created on the Fab Four product page were made in a Beatles style, so the product is very flexible for a lot of uses.

CA: I believe you have your own private studio for sampling sounds. Aside from the economic advantage of not having to rent a facility, were there specific ways you outfitted your studio to optimize it for sampling?

DR: Well, it’s not private; it’s also available to outside clients as well as for our own use. One of things that we’ve done there that we find useful is to have all of our recording rooms and control rooms connected to each other, so if we are working in the large Studio 1 for example and find we just can’t get the right sound for a particular instrument, we can set that instrument up in one of the smaller rooms and we don’t have to move our recording setup to that studio. We also have a video hook up so we can see what’s going on in the other studio. In addition, I purchased a lot of very rare, great-sounding equipment that gives us an almost infinite palette of audio colors to work with. So, we can literally create any kind of sound that we want.

CA: People still sample, and EQ readers are more likely to be among those who create their own sounds. Do you have any tips or advice you could give on how to record high-quality samples? For example, do you feel it makes a difference to record at 96kHz even if the end result will play back at 44.1kHz? Do you prefer to record dry, or with some ambience? Any insights would be appreciated.

DR: We record a lot of instruments in 88.2kHz, not 96kHz—because that way you can halve it if necessary. In digital, math comes into the equation. And it’s much easier in math to divide by two than it is to change 96k into 44.1k. 96k to 44.1k means that the computer is making a lot of decisions that we don’t want. Computers still have to compute 88.2k to 44.1k—but it’s much, much easier to halve the sampling rate. A lot of our current products are released 24-bit/44.1k simply because, if we released them 88.2k, they just wouldn’t play back on many people’s computers. We record them at the higher sample rate for the future. If computers double or triple in power over the next few years which we expect, we can release higher quality versions of what we have already, without having to re-record anything. So, typically, what we do, we record everything 88.2k, we edit everything 88.2k—then we bring it down to 44.1k for release. We very meticulously listen to the conversion to make sure there’s no major quality loss. But there is a subtle difference. For example, Fab Four was released 88.2k because the instruments weren’t as demanding as our orchestral instruments, and I think you can hear it… the transients on the drums are really crisp (love the drums in that collection). So, that’s one bit of advice I’d give.

The real work in making any kind of sampling product is really capturing it right in the first place. And particularly when you go and hire a concert hall and orchestra, things like that, it’s very expensive as well—sometimes hundreds of thousands of dollars. So you don’t want to set up a situation where you have to do it again just because new technology comes along and you want to take advantage of it. So we typically make sure that we have everything recorded for future developments in technology, as well as the products that we’re releasing now. For example, the orchestra that we recorded in a concert hall, we recorded 88.2kHz and 176.4kHz—because you just don’t know what’s coming. If we look back and see where we’ve come in the last 10 or 15 years and then project that forward 10 or 15 years, it’s quite conceivable that we can release something with that kind of sampling frequency—which will obviously give you the most superb quality possible. So, we have that. And we’re retaining that. But we just can’t use it right now because computers already have a hard enough time playing back samples with the big instruments as it is. And it gets to the point where you have to have a serious computer to play anything, and that would just limit our market too much. There are always compromises you have to make with technology, but we don’t make compromises in the recording process.

I also don’t like recording anything too dry. I don’t think instruments sound good when they’re put in too dry a space. You put an instrument in a near anechoic environment it just doesn’t sound like the instrument should, the sound most people identify with. We’re all about making things sound real. Some companies record instruments in a dry environment, and then you have to put artificial reverb on it to make it sound listenable. So now you’ve put two artificial processes on it, so how could it possibly sound real with all of that? We don’t believe in that. Most of our products have been recorded with controllable ambience. That’s what we did with the orchestra project—we had the ambience recorded separate so the composer could control the ambience to suit their composition. And I would argue that, in the case of an orchestra, that 50% of the sound that people identify with is the hall sound. Obviously that percentage changes depending on where you’re sitting in the concert hall. If you’re sitting in the front it might be a little less, if you’re sitting in the back it might be a little more. But it’s still a big part of the sound that the people identify with. And so to then take those players and those instruments to some dry environment and record them with a microphone stuck down the neck of the instrument just seems crazy to me—I never, ever got that.

CA: We all know that software theft has hurt the bottom line for music software companies. But do you think that software theft has also had a slowing effect on the advancement of technology? In other words, if this degree of theft didn’t exist, would our technology be further along than it is today?

DR: Piracy is just something that just really gets to me. We had to put up with it for the first 15 years that we had the company and there was nothing we could do about it. And only when we moved to the computer could we get a handle on it. Our use of the iLok system has proven to be very effective. I’ve yet to see any PLAY cracks out there. People have advertised them but we’ve followed through and found that it was just some shyster taking people’s money and giving nothing in return. A lot of companies suffer from their products being stolen by thieves.

I actually don’t like the word pirate; I think it sounds too romantic. They are common thieves, that’s the only description that is applicable. And I think that at the end of the day everyone suffers from it because there’s less money coming into the system that’s required to make even better products. It’s almost destroyed the record business. I would like to think that people that call themselves musicians or artists have a conscience. They tend to be the poets of our generation and I have a hard time figuring out how they can justify stealing someone else’s hard work in order to create their own.

CA: What is the biggest limitation in virtual instruments that you hope to address in the future?

Mainly the control architecture—we continue to try and refine that, which is another reason for controlling our own software. What we’re trying to do is literally read the mind of the player. What is the player’s intention? When they hit those keys—what do they hear in their head that they want to come out of the speakers? And, we still have fairly primitive tools to do that with. We’re still dealing with MIDI, which is not the most sophisticated system available, but it’s all we have. And I think that what you’re going to see in the future is you’re going to see a combination of samples and synthesis working together. We now have physical modeling. That’s where they create an electronic version of the sound wave. It has a lot of advantages in terms of control architecture. But the actual sounds still don’t sound as good as samples in my opinion, and until they do I don’t think you’re going to see much change in the products that people typically use to produce their work. It’s an area where others and we are doing research, but it’s probably a few more years before you’ll see any products that we’re happy with.

Currently we offer multiple dynamics, interval sampling on some products (SILK for example), different articulations, keyswitches to control them, mod wheel control, and things like that. But analyzing what’s coming out of the keyboard and interpreting that into playing the right articulations is obviously the ultimate goal because we firmly believe that musicians need to get the tech side out of the equation and concentrate on making music. And sometimes, they spend far too much time grappling with the technology—which I think is a bad thing. And it’s up to companies like us to find ways to reduce that; certainly we’ve come a long way to achieving that objective.


CA: Of course, Reason is a complete studio application. But I wonder if you have a sense of how many people primarily use it to ReWire into a host application as a sort of “outboard instrument rack”? Is this a common usage of Reason?

ENB: Yes, it is. Our user surveys have told us that the two most common types of users are those who say Reason is their main (or only) “axe,“ and those who use it with a ReWire host. It fits with our observations too. It seems like either you create your music in Reason—using our sequencer, or you use the rack, sort of as as a super plug-in providing all your keyboard sounds, loops, and the like.

CA: Several companies, like Native Instruments and Arturia, are getting into hardware controllers to provide more of a “hands-on” feel with virtual instruments. And you’ve made an effort to make Reason as compatible with as many control surfaces as possible. Do you feel the inherent lack of physical control, as opposed to a traditional hardware synthesizer, is a limitation with today’s virtual instruments?

ENB: Absolutely! It depends a little bit on where in the creative flow you are, but often just grabbing a knob and twisting it is the fastest and most direct way to take control of your sound. And then a mouse doesn’t really cut it. Generally, with many software designs there is too much stuff that gets in the way of the things you want to do 90% of the times. Traditional hardware designs, like analog synthesizers, are limited but have that accessibility. As with everything else in life, it’s a trade-off.

CA: It seems Reason 3.0 made a conscious effort to make the program more “performance-friendly.” Do you think a lot of your users take a laptop to a gig with Reason serving as a virtual instrument rack, or do you think most musicians aren’t quite ready yet to take a laptop on the road?

ENB: No, they’re fearless! We had a visit from the keyboard player of one of the biggest acts in the world and he uses only Reason and one other keyboard on stage. I sort of assumed he had backup systems and spares of everything, but no, one laptop was all he carried! When we went to see the gig he threw his wireless MIDI keyboard up in the air and didn’t catch it on the way down. He had to finish the gig on the keys that still worked . . . but yes, we do see more an more people just bringing a laptop and Reason. It works, Reason is so solid so as long as your computer holds up you’re fine.

CA: What do you feel are the most common misconceptions about virtual instruments, especially those in Reason?

ENB: The quality of the sound! There are so many misunderstandings about computer-generated sound. It’s hard stuff, so it’s not surprising but it’s still frustrating. For starters, there is no reason why a virtual instrument shouldn’t sound as good as a digital hardware unit. Actually, the computer version has potential to sound better in some ways. There are a lot of things that are hard to get right in digital designs, but unless you’re educated on the subject it’s really hard to know where to look for the differences in quality between various products. We do measurements, A/B tests, blind tests, etc. Sometimes units that we feel are really sub-standard get rave reviews and sometimes we or someone else gets criticized because of a misconception. I really wish someone would debunk a lot of the myths that surround digital audio in music production. It would help both users and the industry.

CA: Some people wish that Reason did audio recording, but I’ve always felt ReWire was a better approach, because it let you use whatever host you like for audio recording. Was this the main reason you developed ReWire, or were there other reasons as well?

ENB: Actually, the main reason was to get ReBirth to work with Cubase. This was back in the old days, even before Reason, remember? But we did it like we always do, we kept it open-ended and gradually opened it up in both ends to where it is now. But the main reason was the same as with everything we do—we just want to help people make more and better music. Hooking up our software synthesizers with a recording application was jut the natural way to go.

CA: Thor is an extremely ambitious instrument. How long did it take to develop, and how many people were involved in creating it? In a related question, what is the process of creating a virtual instrument for Reason? Is it all done in-house, do you have outside people working on it, or what?

ENB: Mats Karlöf was the product manager. Peter Jubel did all the signal processing. The two of them basically designed it together, with input from a lot of other people. Then there was graphic design, developers who worked on the GUI and integration into Reason, sound developers (headed up by Kristoffer Wallman) and so forth. I think it ran over the course of a year or so. Normally we do all the design and development work in-house. Patches and sounds are mainly done by external people. The process of making something like Thor is long and convoluted. It passes through a number of stages of designs and prototypes and ends with user tests and tweaks. It’s really hard to summarize, actually it would be easy to fill up an article on that subject only.

CA: You have now created two big standards for the industry, ReWire and the REX file format. What other standards would you like to see created that involve virtual instruments or recording?

ENB: Wow, I don’t know . . . don’t we have enough standards as it is? Wouldn’t fewer standards be a better way to go? But seriously, better integration between hardware and software would be a good thing.

CA: You’ve taken an interesting approach to copy protection, where you reward legitimate users with lots of free goodies and incentives instead of concentrating on trying to penalize those who steal software. Looking back, do you think this has been a good decision?

ENB: Actually all our main products have copy protection and always have had. Really strong ones too. With ReBirth you had to put the CD in the drive every time you ran the application. With Reason, it’s just that the protection is based on the fact that no one could download 600MB, which was true in 1999 when we created it. But regardless of that, protection is like the lock on the door of your restaurant. It may keep burglars out but it doesn’t bring any customers in. We want the relation to our users to begin when they buy the box, not end there. So we try to come up with little things that make it worthwhile to be part of what’s going on in Propellerheadland. It’s nothing spectacular, it just seems to make sense.

CA: The Reason Instruments series is outstanding. Can you reveal any special techniques you used in the sampling process that explain for why the instruments are so playable?

ENB: Musicianship combined with meticulous engineering. Kristoffer Wallman works with a team of excellent people and guides them towards those results. But I’m glad you say “playable” because again that’s exactly what it’s about for us. Not doing the most advanced or biggest or cleanest instruments, or whatever. Just the most playable, and the ones that work in real life productions.

CA: What do you see as the biggest weakness in the current generation of virtual instruments?

ENB: The integration with the host. And then I refer to plug-ins of course, not Reason. It’s gone little bit haywire I think, where some instruments are more like applications themselves. If your main goal is to write and produce a song, having to shift between so many “worlds,” that are not integrated, not share user interface conventions, don’t allow you to cross-browse through patches etc, it slows you down, it deters you from the original goal. Music making shouldn’t have to be like that.


CA: Digidesign is known for, among other things, creating hardware interfaces for their software. How important is hardware control for virtual instruments, and do you anticipate more hardware solutions appearing along the lines of Native Instruments’ Kore?

PG: Hardware control is very important for using instruments in composing/arranging, not necessarily for programming sounds. I think though there is less and less need—or let’s call it “willingness to pay”—for customized remote hardwares for virtual instruments. For now, remote surfaces and remote keyboards do too good a job. Then, hardware has the unpleasant habit of not magically changing the user interface when you load up a new instrument, so it’s deemed to be very limited across the range of virtual instruments pretty much everybody has these days. Besides, there’s a solution on the horizon that can combine hardware touch and feel with unlimited adaptability to GUIs: Multitouch surfaces.

CA: Many analog purists feel that virtual instruments don’t deliver a true analog sound. Is that still true, or a holdover from the early days of virtual instruments, when the technology was less sophisticated?

PG: I think one reason why many purists feel that virtual instruments don’t deliver a true analog sound is because many of them actually don’t, although the manufacturer’s marketing, reviewers, and public forums state the opposite. I can understand that purists—or maybe just professional musicians who used the real thing in earlier times—after a few disappointing experiences lose faith in virtual instruments as a replacement for analog synths. I am convinced that we have the technology to create a 100% realistic emulation of, say, a Moog 55, but there are remaining practical challenges.

First, those original synths required a pretty good synth designer and selected parts in the first place. All those electronic parts, when connected, interact in so many ways that it’s sure not as easy as hooking up an oscilloscope to the original and analyzing what happens. It requires an extremely good ear and tremendous synthesizer expertise to even know what you’re supposed to emulate, and it’s infinitely complex. So even assuming you got that right (which in reality is very rare), there’s the challenge of computing power. You want to be sample-accurate at least, as an analog synth doesn’t know any buffer sizes. Most commercial products are not sample-accurate, which in essence means the timing resolution of (for example) the envelopes or modulation options is finite, and you can hear that. Many avoid audio rate modulation too, as it eats up a tremendous amount of CPU. Then how about the artifacts created by aged key contacts, temperature drift, interaction of two mod sources fed into one mod destination—all that clipping, voltage/signal compression, beating, all those tiny inaccuracies have to be accurately modeled.

All commercial products have to compromise on these things in a big way, and a purist recognizes that. And even if you got that right, there’s psychology: People who know they’re not sitting in front of the “real thing” don’t want to believe the emulation is not accurate, and sitting in front of a big modular system speaks to multiple senses and has nothing to do with fumbling on tiny pixeled knobs on a computer screen. I bought a Moog 55 replica last year and love it more every day. Sure, it’s 170 pounds for one monophonic voice, but that voice is worth it.

That all said, I still think all those virtual synth emulations have their right to exist—the huge majority of the users of those virtual instruments love them and would never bother or be able to afford buying the real thing.

CA: Can you describe the process of designing a virtual instrument? Is it a team effort, or mostly the result of one person? About how many ma n-hours does it take to develop a virtual instrument? How much effort goes into creating a graphic interface?

PG: I think this differs greatly from company to company. I know companies where one single guy makes all the difference—or even all the products, and I know others who have no one particularly over-talented, but a great team. It also differs from product type to product type. I’d say a sampler is a typical team product because it’s mostly about efficiency, feature set, compatibility and stability. That said, a novel breed of virtual instrument requires one or two people with a lot of background, golden ears and a good idea of what customers will appreciate. The bread-and-butter coding work could even be contracted out—there are so many people out there with a decent knowledge of DSP programming, C++, some GUI framework and plug-in formats.

In our case, the development of a virtual instrument takes somewhere between six months and one year, with a team of about 12 people involved—but they’re usually also something that distracts them going on at the same time, such as Apple feeling like releasing another OS that breaks your plug-ins, or a new Pro Tools release. For a specific example, the five new instruments and 20 effects in Pro Tools 8 were done in less than five months, but of course we could also re-use a lot of existing engine technology. There’s a bit of Structure in the Mini Grand, and a bit of Transfuser in Boom.

Being part of Digidesign, quite a substantial part of our time goes into testing. The graphic user interface usually gets developed along with everything else and rarely gets finished before everything else. Of course we usually draft everything as part of the specification, but in every case when you look at the first prototype, things that made sense on paper or in Photoshop don’t work that well, and you have to change on the fly. I’d say it typically takes three months of one talented graphic designer and a good product designer to develop the GUI. The actual implementation work is not quite that much if you use a good GUI framework.

CA: Are you looking forward to the increased RAM that 64-bit operating systems provide to a system? Does that have uses for virtual instruments other than samplers?

PG: Yes, definitely—it has a major effect on your overall workflow, as every project (at least in my case) uses several instruments at a time. Pretty much every instrument except for a pure virtual analog synthesizer requires RAM of some sort, and that adds up quickly. All sampled real-world sounds rely on disk-streaming—drums, pianos, not to mention big orchestra setups, but also sample-based techno synthesizers—everything. Once RAM isn’t an issue anymore, we’ll probably see more instruments that liberally use huge samples for new synthesis technologies, and companies can stop wasting their precious time on looping and compressing samples. Whether it’s 64-bit systems or solid state drives becoming cheaper and more popular—disk streaming is going to go away real soon, and that is a great thing.

CA: What is the biggest weakness in the current generation of virtual instruments?

PG: A lack of innovation to the degree of annoyance, and it’s so bad that it creates “marketing by desperation” as I call it, where companies seem to lose a healthy sense of embarrassment in their advertising and product strategy. If aliens would read press releases for another me-too virtual grand piano or analog synth emulation, they’d be led to believe we humans have invented the self-powering light bulb. And that in spite of most instruments you see these days being nothing but a regurgitation of other existing ones. And I don’t say “all” merely for the reason I may have overlooked one, but I wouldn’t know any single instrument in the last two to three years that does something I didn’t have available five years ago. I want to be clear: Regurgitation and iteration is fine and in the best case it yields improvement in the bread and butter domain, but that’s all there is right now.

Look at what gets awards these days, and you know those poor juries must be desperate. Of course, every company claims that their newest virtual instrument is the best, total unique, a revolution and creates sounds unheard before—I’ve seen that in virtual grand piano marketing blah. But the only real reason they say that I guess is because “XYZ is a regurgitation of A, B and C” or “it’s not even better than A, B and C but we had a gap in our catalog” or “We could have done this fantastic DFG but let’s face it, you herd of sheep wouldn’t buy it” wouldn’t sound that appealing.

What I find dangerous about this is that it has a dulling effect on customers—over the last three or four years I’ve experienced a decreasing interest in innovative concepts. That’s a catch-22—if your finance guy tells you the best-selling instruments for years and years have been drum modules—with the main development over the years being bigger libraries and lower prices—why bother developing a neuronal atmosphere resynthesis system with GPS interference modeling? And yet there are still unresolved areas that virtual instruments would be great at, and would be appreciated and purchased by customers—it’s just too easy to sample another set of drums.

CA: There seems to be two types of philosophies about virtual instruments: Emulating vintage synthesizers and drum machines, or creating instruments that are possible only in the virtual world. Your new instruments for Pro Tools seem to combine those two philosophies. Was this deliberate? Can you comment more on your philosophy of creating virtual instruments?

PG: Well, firstly there’s my personal philosophy of creating virtual instruments, and then there’s my professional philosophy of the guy being in charge for Digidesign’s virtual instruments.

I personally would spend all my time on inventing stuff that only a fraction of our user base would probably appreciate. Under my Digi hat, my main interest is to listen and cater to the needs of Pro Tools customers. That’s why we did Structure, because everyone from hobby guitarist to Skywalker pro production facility needs a sampler. That product didn’t overstrain my creativity, but it was interesting to find out how much more efficient, easy to use or powerful we can make a product that you can buy from ten other companies. Transfuser was the opposite—it started out as a paper I wrote, and the entire team was so enthusiastic about that instrument that it ended up as a creative effort by everyone. And although on the surface it looks like we just regurgitated other loop-based software, if you dig a bit deeper Transfuser has a lot of great stuff you don’t find anywhere else—like the M.A.R.I.O. algorithmic feature.

The philosophy behind the Pro Tools 8 instruments and plug-ins was pretty simple: There are always instruments that people keep asking for and you’d love to do them, but you know they wouldn’t be a huge commercial success and the economics wouldn’t justify throwing that team on it for months. So when we were planning PT8 and the idea of bundled instruments came up, it was clear it weren’t going to be just “lite” versions of our for-sale instruments, but instruments that complement those, are fun to make and play with. I think people appreciate that we had big fun giving those instruments a bit of an edge. Vacuum looks and sounds like it was built from a few rusty old parts in a Russian army shed, Boom is an almost ironic homage to vintage drum machines, and the DB-33 has this amazing new rotating speaker emulation technology that everybody overlooks.

But I can’t say we deliberately combined emulated instruments and new virtual ones. We just wanted to make sure that with the given technology and resources we create exactly the set of instruments that’s most useful and fun to people, and complements our for-sale products.


CA: Were you taken by surprise when virtual instruments first appeared, or did you “see it coming?”

NB: Much as I’d like to pose as a visionary, the truth is that their evolution was too gradual for it to have been a surprise when they finally “arrived”—that is, when computers became powerful enough to present a serious alternative to hardware synths and samplers.

If you think about the history, somewhat cheesy soundcard-level synths have been around practically since the beginning of the digital age. Therefore the idea of synths’ interfaces being on a computer screen already wasn’t novel, never mind that their engines ran on their own chips. Creative Labs’ Soundblaster card for PCs is an example.

Even before then, Yamaha had an 8-bit music computer in the early ’80s called the CX5m, with multitimbral 4-operator FM synthesis (based around the same chip later used in several of their instruments, such as the dust-laden FB-01 now resting somewhere in my garage). And by the later ’80s, the Mac’s built-in lo-fi SoundManager software-only sampled instruments introduced the concept of virtual instruments that didn’t need extra hardware—even if they didn’t go as far as proving that concept.

There were many other hints of things to come. Jumping ahead a couple of years, Digidesign’s SampleCell was a serious, professional sampler-on-a-card that really bridged the gap between software and hardware. (Much later Digidesign introduced a short-lived software version, Soft SampleCell.) SampleCell used the host computer as a front end, but it had its own memory and processing onboard.

By the mid-’90s, music companies were licensing their chips to computer card manufacturers. An example of that was the Morningstar MacWavemaker, a version of Kurzweil’s synths that sounded very credible; it used the computer as the front end. And the most likely reason Creative Labs bought E-Mu, a pioneer sampler company, was to put some of its technology in their mass-market products like SoundBlaster.

Yet another example of the computer being a front end for hardware is the very high-end Synclavier “workstation,” which eventually ended up using a Mac as its front end control interface. That leads to another important thing: the Synclavier was an early example of a “workstation”—an all-in-one self-contained music system in one box. Yes, even then the idea of a do-it-all musical tool was appealing, and to this day a lot of hardware instruments include workstation features.

Now, I think most advanced users understood all along that every piece of digital hardware is simply a dedicated computer. And we all knew computers were getting exponentially more powerful with each generation, because we were buying them every couple of years! So it was just a matter of time before a single computer could do it all . . . although that’s turned out to be a moving target, since even today the music software we use can bring a single computer to its knees and a lot of musicians use multiple computers in their rigs.

Anyway, it was hardly a shock when, I think a little before the turn of the century, virtual instruments and software samplers started getting pretty serious. At first the latency made them hard to play in real time, but it didn’t take long for computers to get fast enough to reduce the buffer sizes and take care of that problem.

Nonetheless it did take a while for them to become totally serious. You may remember that V.I.s originally earned a rather dubious reputation, in fact some of the pioneer companies in the late ’90s ended up with arrows in their backs. There were undoubtedly business reasons for their demise, but also these systems just weren’t stable enough for studio use, let alone live performance. This wasn’t so much because the hardware was being brought to its knees—which it was—but rather because of the way these instruments interacted with other software running on the same machine. Oy!

That’s all water under the bridge now, and musicians all over the world now use software instruments for high-pressure projects without a second thought, both in the studio and on stage.

Now, what did take me and a lot of other musicians by surprise is how streaming technology turned sampling into a whole new musical medium. Nemesys GigaSampler, the first streaming sampler, read samples off a hard drive rather than loading them into RAM. That allowed samples of unlimited length, and to whet the appetite Giga came with just one single huge (by the standards of the day) instrument: GigaPiano.

Cool. But the latency made it little more than a novelty . . . at first.

Then they sorted out the latency, and a whole cottage industry of sample developers and specialized Windows computer builders appeared. For studio use, hardware samplers started looking pretty limited. Freed from the limitation of holding the entire programs in RAM, Giga’s unlimited recording time allowed instruments to be sampled in great detail. That meant you could have dozens of articulations for each instrument, and the whole field of MIDI programming was born. It was mostly an orchestral technique—dividing MIDI notes between tracks assigned to, say, samples of different string performance techniques—but that expanded to include all kinds of instruments.

Meanwhile most of the interesting synthesizers were and are being developed in software rather than hardware. The one I personally find the most fun is Native Instruments’ FM7 (now FM8). It’s by no means the most interesting synth on the market—I don’t know if it’s even possible to pick a single one that is—but I get excited because it can read Yamaha DX7 voices directly; my first synth was a Yamaha DX7 back in 1985.

CA: Do you feel that hardware synths will eventually be replaced by plug-ins running on a laptop or a hardware host like Receptor?

NB: I think by now the dust has settled, and the way it is today is pretty much the way things are. Racked or well-protected standard computers, laptops, and devices like the Receptor are all being used for live performances alongside hardware instruments.

Each of those has its advantages. Racked or well-protected desktop machines are very durable, and they can hold extra hard drives internally—both for backup and for holding large sample libraries. Laptops are smaller and more convenient, but they’re not as durable as regular machines and lack the internal drive storage space. The Receptor family is durable, very quick to start in the event of a crash, and you can load programs without a monitor if you want.

If I were a touring musician, I’d want a redundant computer system fired up and ready to go in case something went wrong. But then I’d want an extra one of each hardware synth I was bringing too, because things tend to get thrown around. In the studio, most of musicians I know (who tend to be composers) are all using more software.

CA: What’s the biggest advantage hardware synths have over virtual instruments?

NB: The sound of the specific instrument. Case in point: I’m still madly in love with the Yamaha VL1 Acoustic Modeling synth, which came out in 1994. One of the instruments I play is the Akai EWI wind controller, and until recently the VL1 was the only instrument I used with it. Now there’s a softsynth, Wallander WIVI, that I also use with the EWI. More generically, hardware synths are self-contained systems and everything is more immediate. Turn the thing on and play.

CA: What’s the biggest advantage virtual instruments have over hardware?

NB: You can run lots and lots of them, the computer screen tends to give them a nicer interface, and you can combine many of them inside one computer and recall huge set-ups with a single command. Ultimate Support doesn’t even make the A-frame keyboard stands we all used in the ’80s and ’90s to accomplish the same thing anymore!

The other advantage is that V.I.s are far less expensive, as they run on general-purpose computers; that itself creates an economy of scale. I mean, every Mac comes with GarageBand pre-installed, and that has a whole bunch of instruments and effects that aren’t at all shabby. And all the sequencing programs come with a healthy selection of built-in instruments.

Finally, it used to require a big company to develop a synth, with all the hardware and custom chips. Now there are people in their bedrooms developing software instruments.

CA: What obstacles and weaknesses need to be overcome with virtual instruments?

NB: I think soft synths have pretty much been figured out. Software sampling doesn’t have obstacles or weaknesses, but we’re just now approaching the point where the software can make use of all the available horsepower and memory in today’s top-end computers. Memory comes down to having 64-bit memory access (already there with some samplers, such as VSL’s Vienna Instruments 3 on Windows, soon to follow when Mac OS X 10.6 makes that possible). Meanwhile, there are ways of accessing lots of RAM on Macs by running multiple stand-alone instances of the sample players. Both 64-bit memory access and multithreading are well underway.

I’m not sure the single machine solution will ever arrive, though, because developers are always pushing the envelope. So musicians running huge projects will probably continue to employ slave computers that do nothing but play sounds—in other words they behave just like external MIDI modules.

With that in mind, the next great development is arriving quickly: audio and MIDI running over Ethernet, so you don’t need audio and MIDI interfaces on your slaves. VSL’s Vienna Ensemble 3 does that very well for their libraries, Audio Impressions Audioport runs audio between Windows and Macs, MusicLab Midi Over Lan runs MIDI between Macs and/or Windows machines . . . we’re almost there with both audio and MIDI running freely between Macs and/or Windows for every virtual instrument, regardless of developer.

CA: How do you feel about emulation of vintage instruments as opposed to creating new types of instruments (e.g., Cakewalk Rapture)?

NB: It’s all good as far as I’m concerned. Bear in mind that since my first synth was a DX7, it’s not terribly important to me whether a recreation of an analog synth that came before it is exact. What’s important to me is that it sound really good; other people who started earlier are wont to disagree vehemently.

On a personal note, I find Native Instruments’ FM7 and FM8, which start by emulating the DX7 and then take off from there, just too cool. There are three DX7 sounds I programmed back in the ’80s that I still use, and being able to play them in software is . . . well, there’s something reassuring about having access to the same synth I relied on for everything 25 years ago.

CA: Do you feel hardware controllers for synths are essential, unnecessary, or somewhere in between? Does NI’s Kore system address that need well?

NB: They can be good for performing sounds in real time, sure. I personally don’t feel that something’s missing in the art of synthesis without one, but a lot of synthesists really value the tactile response.

What I really like to use is a breath controller, because it adds a human, expressive feel to parts when you play them. Disclaimer: I’m involved in developing a higher-end breath controller as an alternative to the only one on the market now, the Yamaha BC3. But the reason we’re developing it is that it’s something we want to use.

CA: What are your favorite virtual instruments, and why?

NB: Because of my personal musical interests, my favorite instruments are all sample libraries. If I had to pick one library for the desert island, it would have to be one of the major orchestral ones on the market, each of which has its relative advantages and disadvantages. As to synthesizers, the sounds that inspire me the most tend to be instruments that could exist in the acoustic realm but don’t.

CA: Do you have an idea of what the next step is beyond virtual instruments?

NB: Whatever it is, I expect the next step to be within the field of virtual instruments, not beyond it.

If computers continue to get way more powerful, one could imagine that convolution processing would take on an increasing role. Right now we’re using a single sample of whatever is being convolved, but at some point it should become possible to go between multiple impulses in real time. VSL is already doing that with their MIR mixing/reverb processor, but I could see that being applied to real-time synthesis as well. And for that matter, to sampling.


Applied Acoustics
Future Audio Workshop
IK Multimedia
Native Instruments
Propellerhead Software
Sonic Core
TC Electronic
Use Audio
Way Out Ware