The sounds are vastly complex and produced, and the unit's humanistic formant-shaping-synthesis technology is fascinating and different. Dig beneath the

The sounds are vastly complex and produced, and the unit's humanistic formant-shaping-synthesis technology is fascinating and different. Dig beneath the presets, and you'll find yourself in an extremely dense and complicated operating system, immersed in operators and algorithms. This is territory that will be simultaneously familiar yet unfamiliar to all the players around the planet who bought the world's second-best-selling synth back in the 1980s. Let me present to you the DX7 2000 in a box, aka the Yamaha FS1r.

If you still have an old FM synth but curse its now rather lo-fi output and plain-Jane patches, you will find that the FS1r is to die for. You can transfer DX patches into the FS1r and give them a fresh coat of paint by applying new filters and effects. That will put you several years ahead of most FS1r owners, who will definitely spend their first year with the instrument playing presets and steering well clear of the edit buttons.

ON THE CASEYamaha has not gone overboard with the hardware. The FS1r is a standard 1U rack-mount unit that offers 4 free-flowing control knobs, 15 teeny mode/access buttons, and a good-size (3- 5 1-inch) backlit LCD. The power switch, headphone socket, and master volume are also front-mounted. At the back are the main L/R outputs, a set of audio outputs (Part/Voice), and MIDI In/Out/Thru (see Fig. 1).

The four control knobs have three separate functions. When you press the upper of the two LED buttons, they serve as global tweakers of Attack, Release, Formant, and FM. Press the lower button, and the knobs become four assignable controllers. In Utility mode, they provide direct access to Part/Op, Group, Cursor, and Value.

I'm grateful for any immediate, real-time control over sounds within a rack module, but to some extent, the knobs' power is restricted by the almost unfathomable number of options available. If you leave well enough alone and allow each Performance to be controlled by whatever assignments the Yamaha programmers decided would be cool, fine. But start tweaking, and the complexity of choosing options soon discourages you from investigating them. You might also encounter some anomalies that I'll detail later on.

In general, I found the user interface pretty grueling; it's reminiscent of Korg's Wavestation SR. Luckily, as with the Wavestation, the preset sounds and available features are such that you can learn a little, forgive a lot, and move on.

THE MAIN DEALThis is a turbocharged, 32-note polyphonic FM tone generator. With a maximum multitimbrality of four parts, it's clearly not designed as an all-things-to-all-people module. GM it's not, though it does seem to respond to Yamaha's XG format in terms of NRPN control.

The FS1r sounds different. It sounds new. It includes some devastating natural-instrument sounds, particularly in the electric-piano and bass class, but it also ventures into the uncharted waters of formant sequencing and real-time control over timbral content. The response is superfast, which makes it a perfectly acceptable solution for a gigging player who admires the classic nature of FM pianos but wants to use them in a cooler, more modern environment. And there's enough under the hood to keep even the most ardent programmer merrily tied up in knots for years.

You can edit from the front panel, but Yamaha also has a Mac-only editor available for downloading on its British Web site (www.yamaha.co.uk/synth). Frankly, neither approach is remotely friendly. Documentation for the instrument itself will make no sense to 99 percent of the buying public, and documentation for the editor software is nonexistent.

As I said: DX7 2000, welcome back!

BASIC THEORYThe FS1r sells itself on FS (formant shaping) synthesis, which, as Yamaha admits, is closely related to the company's groundbreaking FM (frequency modulation) synthesis in the DX series. If you understand FM, you will understand the principles here, even though the range of parameters, operators, algorithms, and processing options has been increased substantially. (For an overview of FM synthesis, see "Square One: FM Basic Training" in the April 1999 issue of EM.)

If you don't understand FM, you will be joining the 199,900 DX7 owners (out of 200,000) who had no clue how to program the DX7 from an initialized patch. It's the same with the FS1r, thanks to the fiddly user interface, tiny panel buttons, and a zillion levels to each page. Adding to the problem is the fact that this is a rack-mount module, and as with most such modules, it's almost impossible to view the screen from a comfortable angle. A centralized Value knob would have been helpful.

The word formant in formant-shaping synthesis refers to the defining timbral characteristics of vocal sounds. Certain fixed frequencies are amplified by the vocal tract, whereas others are attenuated; the amplified frequencies are called formants. With the FS1r, these timbral characteristics can be applied to other types of sounds as well.

Rather than using a conventional system of filtered oscillators and noise generators to imitate consonants and the various subtle changes that make up complex vowels (ai, ee, ow, and so on), Yamaha turned to its trusty FM synthesis system. In a slight twist on the old system, both pitched and unpitched operators (each of which includes an oscillator and filter in one building block) can be ordered in a variety of ways within 88 different algorithms (see Fig. 2).

As in FM of old, the operators can be either carriers or modulators, but unlike old FM, FS operators can be either voiced or unvoiced. A voiced operator is capable of producing one of eight waveforms: sine, broadband with all harmonics, narrowband with all harmonics, broadband with odd harmonics only, narrowband with odd harmonics only, resonant broadband, resonant narrowband, and formant. The first seven of these waveforms can vary in pitch as different notes are played, or they can remain at a fixed frequency; the formant waveform can only be a fixed frequency. In addition, the user can specify the bandwidth of these waveforms. Unvoiced operators are primarily for noise generation.

Add to the mix an elaborate matrix of envelope generators, separate filters, and LFOs, and an intriguing time-based feature called Formant Sequences (FSeqs), and you have yourself a hefty bag of programming tools.

But how do they all manifest themselves? Let's take a look.

PERFORMANCE ANIMALWhen you power up the FS1r, the unit defaults to Performance mode. Flip through its 512 Performances (384 preset and 128 user-definable), and you'll go from classic DX7-type pianos (on steroids) to swirly pads, gnarly organs, and basses to radical vocalized patches that actually seem to speak to you.

You can play these sounds or tweak the four front-panel knobs to make slight changes. Just about all you want to see is on the main screen: patch number, pan position, volume, reverb and effects levels, and key transposition. Although you'll face rather more error than trial when accessing these day-to-day parameters, you'll be very happy.

The FS1r's highest level of operation is a Performance, which is made up of a maximum of four Parts (hence, the four-part multitimbrality maximum). Each Part can operate on its own MIDI channel, and it has its own Voice and control assignments, such as filter, EGs, pitch, portamento, and note range.

A Voice is the kernel of the sound, where you assign the algorithm for the operators. Rounding out the Voice parameters are filtering, LFO, and envelope-generator modulation. You can save some 1,500 Voices in memory (most of them will be preset Voices in ROM, but you can save 128 of your own Voices in RAM). Effects and EQ, you ask? They're applied at the Performance level.

You might well decide that this is simply too many layers of control. Where is filter resonance applied? Within the Part? At the Voice level? Within a particular algorithm? Often, it's not easy to figure out.

Knowing the FS1r's FM heritage, you can't help but dive straight into its collection of electric pianos. "EP Wide" is as creamy a digital Rhodes as you're going to find, with a beautiful Velocity response. It's got depth but plenty of cut; it's rounded, but it has a natural attack. The patch just plays itself.

The FS1r also has a superb accordion; a majestic church organ that'll have you on your knees in seconds; plenty of speaker-rattling, sine wave, drum 'n' bass-type basses; enough funky vamping patches to keep gigging keyboard players afloat for the next couple of years; and a seemingly endless array of, well, weirdness.

As all sensible manufacturers do these days, Yamaha provides a little onboard search engine so you can scroll through sounds by category. This is available for Performances and Voices, and you can add your own creations to the lists.

In summary, even if you don't tweak a darned thing, the FS1r is a sonic monster.

SO TO TWEAKArmed with four substantial front-panel knobs, even the greatest preset hound will soon fancy a tweak. The first two knobs, Attack and Release, globally alter the envelopes in the whole sound (that is, all the Parts and Voices within a Performance). But what of the Formant and FM knobs?

Well, the documentation on these knobs is nothing short of a travesty. Essentially, the manual says that increasing the level of the formant increases the level of the formant, and the same for FM. Great, but what does that mean? In practice, these knobs add harmonics, boost resonance, and lend weirdness to the sounds. Within each Voice, you can set up five operators for formant and FM control, each with its own range of values and destination. The complexities of FM being what they are, results can be very unpredictable. When you consider that a Performance can use several Voices at once, you begin to see why the authors of the FS1r manual left so much to the imagination.

Perhaps it would be best to describe the knobs' effect on a specific patch. With the EP Wide patch, the FM knob thins and sharpens the sound to the point where it's like a pedal steel (especially if you soften the attack as well). The Formant knob adds an intriguing burble, most notably with increased Velocity. Soften the attack a lot and add a touch of release, and you have one of those dreamy, ghostly piano things you can play all day.

Real-time tweaking is one thing, but what about recording front-panel changes into your sequence? The menus within the unit imply that this is perfectly possible, but I was unable to do it. Fortunately, the FS1r does respond to NRPNs and CCs, which I sent from my PhatBoy MIDI performance controller. Using knobs on the PhatBoy (in XG/GS mode), I recorded lots of filter, envelope, and LFO manipulation into Cubase and everything was recorded and played back just fine.

FORMANT SEQUENCESFormant Sequences (FSeqs) are great fun, but you can't edit them directly on the FS1r (or, I suspect, by any other means). Somewhat similar to Ensoniq's Transwaves and Korg's wave sequences, FSeqs are sequences of formant structures, streams of complex multistage envelope data that can be applied to sounds in order to make them speak. The FS1r has about 90 FSeqs that you can apply to Parts within a Performance. The sequences can be triggered once by the first note played after silence, or repeatedly by all notes (though this option gets too messy for most purposes).

You can set a loop within each FSeq and speed it up or slow it down. For instance, with the ShoobyDoWap preset, ShoobyDoWap can become the sequence ShoobyDoWap, byDoWap, byDoWap, byDoWap or ShoobyDoWap, DoWap, DoWap, DoWap, and so on, all at various speeds, directions, offsets, and delays.

A slightly more useful application is to put an FSeq under Mod Wheel control so that you can access different syllables or sections of the FSeq at any rate in real time. Apply this to a Clavinet-type patch and you'll say "Dead funky!" though in fact, it's very much alive.

The list of FSeq presets includes such memorable epithets as Oiyai, ChowaUu, Thankyou, YaYeYiYo, and CanYouGi, but it's a shame you can't create your own FSeqs directly on the unit. I suspect many would want to train their FS1r to order coffee or say "P-off" when the engineer comes into the studio.

EFFECTSThe FS1r has four separate levels of effects processing: reverb, variation, insertion, and EQ. As you'd expect from Yamaha, the reverbs (halls, rooms, plate, tunnel, canyon, and so on) are ultrahigh-quality, with not so much as a glitch or ping in sight. You can pan the reverb and make microscopic edits to densities, sizes, and amounts.

Variation effects include chorusing, flanging, phasing, gates, compression, and delays-in other words, a full range of multi-effects. Again, all are minutely editable in terms of speeds, amounts, and other salient parameters.

Insertion effects are inserted directly into the signal flow, as opposed to the send/return arrangement used for the FS1r's reverbs and variation effects. They overlap the variation effects, particularly in the chorus, flanger, and delay areas. But they also include distortion, overdrive, rotary speaker, and more, for a total of 41 effects in all.

A final coat of sophisticated 3-band EQ can be applied at the end of the signal chain. This is a very powerful selection of processing tools, all the more heart stopping when you consider the size of the sonic palette available before you even think about effects.

As seductive as they are, effects can be dangerous, and they can clutter up the final sound. So I was pleased to see main-page access to the reverb, variation, and insertion effects from within a Performance.

VOICE EDITDetailed programming, be it from scratch or simply from informed preset tweaking, is a daunting task using just the FS1r. I hope that someone produces a better software editor, or at least some route maps into the editor that Yamaha already offers.

As I've said, prior knowledge of FM will obviously help, but the FS1r goes a lot deeper than that. It offers not only eight voiced and eight unvoiced operators (in contrast with the DX7's six voiced operators), but also many more waveform options, resonant filtering as a separate programming layer, pinpoint-accurate pitch envelopes, two LFOs, and a wealth of modulation sources and destinations. Until Yamaha (or someone else) comes up with an editor that's presented and documented so that the parameters are readily accessible, the FS1r is effectively a tweakable, rather than editable, unit.

Does this matter? Not really. During the time that I spent with the unit, I definitely got $999 worth of song-inspiring patches, cool new tricks and treats, and just the sheer pleasure of with it all.

Julian Colbeck reviewed the very first DX7 back in the early 1980s, direct from David Paitch's studio in Los Angeles. Few instruments have avoided his batlike ears since then, though today he is more at home producing instruments for Keyfax Software & Hardware than examining them.