January's Winter NAMM show in Los Angeles was the biggest on record, and offered a glimpse at most of the new innovations the music industry will have to offer in 1987.

IT'S SAID THAT the fewer distractions an area has, the better a venue for business meetings it is likely to be. If that's true, then the city of Anaheim, in an area south of Los Angeles known as Orange County, must approach the ideal. Because aside from Disneyland, Anaheim offers nothing of any great interest other than a huge convention centre, a clutch of hotels and motels, and all the paraphernalia associated with business gatherings.

No wonder the National Association of Music Merchants (NAMM to you) chose to locate their winter convention there - though if things go on the way they are, Anaheim might be too small for future NAMM conventions.

As many people predicted, Winter NAMM '87 was the biggest ever, both in terms of the number of companies exhibiting, and the number of buyers attending. With the annual cycle of new instrument launches (especially in the hi-tech area) finding its natural start-point at around the time of the huge Frankfurt Music Fair in Europe (see report following this one), it was only a matter of time before the NAMM show that traditionally precedes Frankfurt would grow and grow - as Winter NAMM has done consistently over the past four to five years.

Yet crazily, Winter NAMM still lasts for only three days (Summer NAMM lasts four, Frankfurt five). And big though the Anaheim Convention Centre is, it couldn't hold all the companies that wanted to exhibit at this year's show - which is why NAMM are now scratching their heads, wondering what to do in the couple of years before the Anaheim Centre gets an extension added to it.

As for those companies lucky enough to get stand space at Winter NAMM '87, many found themselves tucked away in unlikely corners and alleyways, as the NAMM authorities endeavoured to make use of every available inch of space. Not surprisingly, there was an above-average number of demonstrations going on in the neighbouring hotels this year, too.

So, what new innovations were all those companies scrambling to get space to exhibit, and what new technological marvels were retail shop owners (not to mention members of the music press) scrambling to get space to look at?

Well, the hi-tech instrument that got the biggest pre-show hype was Yamaha's successor to the DX7 - a machine which has, after all, revolutionised the way we go about creating synthesised sound, and the way the general public hears it.

In a display of utterly slick and predictable professionalism that is typical of the way musical instruments are sold in the States, Yamaha's demonstrators gave us a (possibly) tongue-in-cheek obituary for "this knighted node of the MIDI network", before going on to explain that the presence of two new keyboards ameliorated the death of the DX7 to the extent that its passing was more like "a walk through the digital lounge bar of life".

The two new keyboards in question are designated DX7IID and DX7IIFD. The "D" in the first refers to the fact that, relative to the original DX7, it has "Dual" FM tone generators, as on the DX1 and DX5 synthesisers, both of which had dual six-operator, 32-algorithm synthesiser sections. The "F" in the second refers to the fact that it has a "Floppy" (3.5-inch) disk drive onboard, for storage of voice, performance, and tuning data from the new DX, plus MIDI data from external synths, sequencers and drum machines.

Both new machines are "upwardly compatible" with the old DX7, so that even though they use a larger RAM cartridge with twice the capacity of the original Yamaha one, you can slot your old DX7 cartridges straight in to either of the new keyboards using a special adaptor. Yamaha's demonstrators claimed this would result in "an immediate increase in fidelity", but fidelity to what? To the acoustic sounds your patches are trying to emulate (if any)? To your original conception of what the patches should be like? We were confused.

More encouragingly, though, the new DX7s have a number of new features that should make them more responsive instruments to play and program, and a couple of somewhat off-the-wall capabilities whose appearance on modern synthesisers is refreshing, to say the least.

However, you'll find an exclusive, in-depth review of the second-generation DX7 elsewhere in this issue, and seeing as there are any number of other new instruments to consider, we'll proceed forthwith.

A new implementation of FM technology appeared on the Korg stand, however, just five yards from where musicians and shop owners were falling over themselves to get their hands on the new DX7s. Korg's first all-digital synth is called the DS8, an eight-voice instrument whose internal architecture comprises such familiar components as oscillators (two per voice), and which involves the user in nothing more complex than a three-tier programming system with pitch, timbre and volume as its main components.

What's more, the DS8 is multi-timbral, so you can assign each voice to a different MIDI channel, thereby making full use of a number of different splitting and layering options, and making the synth a powerful ally for any multitrack sequencer.

In keeping with what is now common Korg practice, the DS8 has a patch-programmable digital delay onboard, while its keyboard is sensitive to both velocity and aftertouch, and a "Performance Editor" section of switches and sliders allows you to switch oscillators, turn touch-sensitivity on and off, and adjust other parameters while playing.

Not bad for under £1000, especially when you consider that, subjectively speaking, the DS8 has the kind of sparkling clarity that is one of FM synthesis' strongest selling-points. Yet it should be no more difficult to program than, say, Korg's own DW8000 synth, which uses altogether more conventional technology to produce its sound. So, good news for synth players who like the FM sound, but who've shyed away from it in the past because they've found FM programming systems unfriendly.

YET IN MANY ways, innovations such as the DS8 were very much the exception that proved the rule at Anaheim this year. Because on many other stands, the theme was one of refining and updating existing ideas and systems. The Yamaha DX7II was probably the example of this phenomenon that aroused the most attention, but other examples of it included the Sequential Prophet 2002 Plus (now with a megabyte of memory and eight separate audio outputs on its back panel) and VS Rack (a Prophet VS in rack-mount form, no less); Ensoniq's cheaper (£995) replacement for the Mirage keyboard (known as the DSK); Korg's range of updates for the DSS1 sampler (now you can increase the memory to 2Megabytes, and halve that excessive load time for samples); and a number of updated signal processing units, such as the Alesis MIDIverb 2 (more on this later) and Ibanez SDR1000+ digital reverbs.

Moving back to the new Mirage, before you convince yourself that you should have waited before buying the old model, take note that some corners (namely at the input filters in the sampling circuit) have been cut, so it's questionable whether the Mirage DSK will sample as well as the original model. The DSK does have stereo outputs, but in all other respects, the Mirage and the DSK are identical. The new model loads any of the old model's disks, and vice versa. Also part of Ensoniq's display was the rackmount version of their ESQ1, the ESQ-M, and a rack version of their digital piano, the SPM1.

For those who already own a Mirage, however, there is hope. Indian Valley Manufacturing, of Telford, Pennsylvania, have introduced the IVM MegaBank Expander, which increases the Mirage's RAM by 1 Megabyte, enabling the updated Mirage to load entire disks at once, and improving the sampler's performance in live situations no end.

Elsewhere, Roland unveiled modular versions of their JX10 (the MKS70), Alpha Juno (the MKS50) and S10 (the MKS 100), as well as taking the wraps off the PR100 digital MIDI sequencer, itself a scaled-down version of the already successful MC500 MicroComposer. The PR100 is aimed at the keyboard player who hasn't yet had much experience of sequencing, and is therefore a good deal simpler to use than its more sophisticated brother (it's only a two-track machine, after all). Yet it has a 17,000-note capacity onboard, plus the ability to store sequence data on 2.8-inch Quick Disk, thanks to an onboard disk drive.

But the PR100 faces competition from the Korg SQ8, which must rank as the cutest sequencer yet designed. It's not much bigger than a packet of cigarettes, yet has an eight-track MIDI sequencer with a 6500-note memory. Like the PR100, the new Korg offers controls similar to those found on a conventional tape recorder, so it shouldn't be too difficult for inexperienced users to start making their first multitrack recordings in real time. Step-time recording is also possible, and the SQ8's LCD is nice and big (considering the unit's overall size), offering information on track status, tempo, current beat and measure, selected MIDI channel, and available memory all at the same time.

The digital drum machine - now getting on for a decade old as a concept - continues to be a popular means of getting a beat down onto tape (among other things), and there were a couple of interesting new machines on show at NAMM. One was the Yamaha RX5, the company's new flagship beat-box. A total of 64 different drum sounds is available initially (more will be available on cartridge as time goes by), 24 of which can be used in any drum pattern. These sounds vary from conventional "rock" drums, through Latin percussion sounds, to DX voices such as marimba and Clavinet. But more excitingly, each drum voice is individually editable over a range of parameters including tuning, envelope, gate time, pitch-bend, looping and reverse. Meanwhile, each drum beat in a pattern is individually editable for pitch, level, envelope and reverse.

You can take things further than this by assigning a single drum sound to the length of a connected MIDI keyboard using a feature called Tunable Note, which gives you 61 notes (if your keyboard is that long) of tuned percussion from a single voice. And that voice is velocity-sensitive, not only to amplitude, but also to envelope attack and decay, so that it changes in timbre - as well as level - the harder you hit the keys. So aside from sounding good, the RX5 is perhaps more likely to sound "human" than any other drum machine - assuming programmers make good use of those editing options.

Less ambitious than the RX5 - though more affordable - is the Korg DDD5. This features the same touch-sensitive keys as the company's upmarket DDD1, and also offers the potential for sound library expansion via ROM and RAM card slots (though there are only two of these, as opposed to four on the DDD1).

But the most interesting thing about the DDD5 is the way it can operate in two separate modes which, strangely enough, make it simpler to use. The first mode allows fully programmable operation for such things as writing patterns, linking them together to produce a song, and so on, while the second allows the same control switches to be used to select preset rhythm patterns, all of which are editable in the first mode. An alternative job command card is all you need to remind you which mode you're in, and with luck, the provision of presets will result in more people getting into programming their own rhythm patterns: give people a head start, and they should be able to do the rest themselves.

Maybe the same will also be true at the other end of the digital drum scale, where Simmons are putting the finishing touches to their new electronic drum system, the SDX. This is an incredibly comprehensive drum sampling (16-bit), processing, sequencing and performance system, but like all good sophisticated systems, it allows the user to enter it on a number of different levels.

Perhaps the most striking feature of the SDX (aside from the arresting appearance of its programming "brain") is its pads, which are 'Zone Intelligent", as Simmons puts it. There are 16 concentric zones across each pad, and depending on which of the three levels you are using the SDX at, you can program these zones so that, for example, you can crossfade from one drum sample to another as you hit different areas across the pad, or trigger one sample at different dynamic levels by doing the same thing. This crossfading technique doesn't just allow the SDX to simulate the action of acoustic drums more accurately; it opens up a new world of percussion control and technique, in much the same way as positional and velocity crossfading have helped transform keyboard samplers from banal regurgitation machines into instruments of genuine creative potential.

AS WE IMPLIED at the start, there were a few debuts made in controlled environments other than those on the floor of the Anaheim Convention Centre.

First of all, Kawai spent the entire show without ever indicating that their new product line (expected availability in April/May) was within walking distance of their stand. Indeed, having a look at their new instruments involved attending a Kawai-sponsored concert (with Steve Smith's Vital Information), having a drink or two, then working your way across the room to where the new goodies were displayed.

Of these, the most significant was the K5, the K3's big brother, capable of real-time additive synthesis, employing six-stage envelopes to control the mix of each of 128 harmonics (!), a digital filter, and final amplifier stages for each of 16 voices. The K5 is multi-timbral, and each of its voices can be set to play over specified note ranges and velocity ranges. The machine also brings back the release velocity-sensitivity the industry has been missing out on since the Prophet T8 was discontinued. The new "K" also features a healthy backlit LCD so the harmonic structure can be observed without relying on terminal support.

Then there's the R50 drum machine, a low-cost version of the R100, which is itself now becoming widely available in Britain, and will shortly be reviewed in these pages. Although its features are limited in comparison to those of the R100, as is its memory capacity, the R50 features the same 12-bit, 32kHz sample-rate drum sounds, and also offers assignable digital effects such as delay, flanging, and gating.

A year ago, the biggest buzz at NAMM was the appearance of a host of easily affordable sampling keyboards. But in 1987, the only really new keyboard sampler was in the Casio room. Their FZ1 was the subject of a great deal of pre-show rumor concerning its 16-bit sampling capabilities. Well, the instrument on show at Anaheim was very much a prototype, and only a handful of sounds were demonstrated for it. Whether it is a genuine 16-bit sampler (with the increase in fidelity over 12-bit systems that implies) or whether it simply uses 16-bit processing (in common with many other keyboards already available) was still to be confirmed by Casio themselves at showtime. The most we could discover is that although the samples Casio have taken for use with the FZ1 are indeed in 16-bit format, the A-to-D conversion for user sampling is still using only 12-bit resolution.

On other stands, sampling technology was being applied in ways other than the ones we're used to seeing on keyboard samplers. But seeing as two of the most significant instruments to show evidence of this were Oberheim's DPX1 sample playback module and the Sequential Studio 440 drum machine/sampler/sequencer (both reviewed in last month's MT), we won't bore you by repeating details on those. Latest news, though, is that Oberheim is about to make an upgrade for separate audio outputs available at extra charge, while Sequential engineers successfully tested the transfer of sample data over the SCSI interface for the first time during the NAMM. That's good news, given that SCSI is finding favour with more and more people within the music industry; you can now get it on the Korg DSS1, too.

In the world of drum sampling, a company called Forat Electronics is supporting all the machines made by the now-defunct Linn concern with worldwide service, modifications and sales of remaining units.

Their latest updates are a MIDI retrofit for the LinnDrum and software due shortly for the Linn 9000, Linn Sequencer and MIDIStudio which implements MIDI Song Position Pointers and fills out the SMPTE option. Hardware modifications include a 320K sampling update for the 9000/MIDIStudio which allows user samples to be placed on all the pads.

Kurzweil were demonstrating their new rack-mounting K250 RMX with their own MIDIBoard controller. The K250 RMX is essentially a rackable version of the company's K250 digital keyboard, and permits sampling at variable sample rates, and also features a 12,000-note, 12-track sequencer.

Other rack modules of note included 360 Systems' new MIDI Bass and Voice Module. The former is basically the same unit as the original MIDI Bass, but now has improved MIDI control, and allows modification of the factory-supplied samples. The Voice Module (subtitled Lead Line Percussion Expander) takes advantage of LSI technology to offer eight digitally-recorded percussion sounds on each sound chip, of which 16 may be installed for access to as many as 128 internal sounds. Like the MIDI Bass, the Voice Module allows samples to be assigned to keyboard zones for flexible control by external MIDI devices.

ON THE CONTROLLER front, Roland were demonstrating the modules listed above with various new MIDI controllers, such as the new GM70 guitar-to-MIDI system. In fact, this turned out to be one of the most convincing demonstrations of a guitar controller we've seen. The converter, which works on any guitar with any gauge strings so long as a Roland GK1 pickup is being used, can determine the fundamental frequency in one half of a cycle.

It is easy to be impressed by specs, and it is easy to overlook details while in the middle of a trade show like NAMM, yet Roland's product specialist, Paul Youngblood, went out of his way to demonstrate how valid this claim is by strumming the guitar furiously, an act which the GM70 appeared to take in its stride, never appearing to glitch at all. In addition to such impressive tracking specs, the GM70's MIDI appears to boast as flexible an implementation as we've ever seen, even on a keyboard controller. Each string is independently programmable with its own MIDI key transpose setting and up to four MIDI channels of your choice, so you could easily program the GM70 for extensive layering and splitting when driving multiple MIDI synths.

Meanwhile Beetle, the US company responsible for the PR7 and QR1 MIDI accessories, surprised us all with their Vortex guitar-to-MIDI converter. The system uses sonar to see how far down the fretboard your fingers are, then determines the frequency of the note you're playing. Theoretically, this should eliminate the conversion delays for which other guitar-to-MIDI conversion systems are so notorious.

The prototype Vortex displayed at the NAMM show was designed for a Fender Strat, and the sonar system mounted over the Strat's fingerboard very tidily. The Beetle-supplied fingerboard carried the words "MIDI Strat" - a concept which is appealing in itself, even if none of us were allowed to play the instrument for ourselves.

Another new stringed MIDI controller to please the crowds at NAMM was the IVL Touchboard MIDI converter system for the Chapman Stick. Demonstrations of the Stick with a Yamaha TX7 showed little or no signs of tracking problems or conversion delays. The IVL Stick-to-MIDI interface unit costs $900 in America, including retrofitted pickups and preamp on the stick. British-based interested parties should contact Rod Argent's Keyboards for more details.

K-Muse had the latest incarnation of the Photon MIDI Converter, with their new Hyperspeed guitar, a Gibson equipped with the Photon pickup, and a bass doing the same (the HyperOctave Bass). K-Muse were also demonstrating their new Foot Controller, which provides the controls needed to access the new software's arpeggiator, sequencer, and hold/transpose functions.

And speaking of infra-red pickups, Audio Optics unveiled their QED pickup, which uses infra-red light to measure both axes of each string's vibration, producing an uncoloured signal since there is no magnetic interference from an ordinary pickup. To prove a point, Audio Optics demonstrated the pickup installed on a Strat set up with nylon strings. Interesting, to say the least.

SOFTWARE COMPANIES WERE well represented at this year's show, with a section of one of the halls dedicated to music software, plus the odd hardware exception.

With the electric guitars, "soundproof" drum booths and valve-amp stacks clear on the other side of the hail, software packages were displayed and explained to interested parties without raising voices - a pleasant change from past shows, in which software companies have been scattered around to compete with drums and guitars for NAMM attendees' attention.

Let's start with packages based around the IBM PC, since Alan Sugar's new Amstrad PC-compatible computers are already starting to take the UK micro market by storm.

One particularly popular IBM-based package at the show came from Performance Computer Concepts: the MIDI Manager 7, an integrated MIDI system available in various configurations for stage or studio applications. The MIDI Manager 7 features a 7x7 MIDI switching matrix, generic patch librarian, 6-track sequencer, and preset handler for co-ordinating multiple synth arrangements. The MIDI Manager is designed to eliminate the need for an assortment of MIDI accessories, making life easier on stage and in the studio. But packages run from $1795 to $2049 in the US, so it's not cheap.

Another musical application for the IBM PC is Music Magic's 16-voice synthesiser card, featuring stereo outputs and built-in 65,000-note sequencer. Four Music Magic synthesiser cards may be installed in the computer for 64-voice capability, and each card costs just $795.

There was also plenty of activity in the Atari ST arena. Passport Designs have come in with MIDISoft, a 32-track sequencer for $99. This is in addition, of course, to their extensive line of Macintosh and IBM sequencing and score-printing programs. Nearby, Sonus were displaying their new Masterpiece sequencer and SuperScore programs for the ST and Macintosh. Masterpiece is a 32-track sequencer, while SuperScore is designed for printout of Masterpiece and Glasstracks (Sonus' budget sequencer for the Mac or ST) sequences, while incorporating its own sequencer and editor.

Sonus also introduced a few hardware items, including the MDM80 MIDI-to-FSK tape sync converter, the MT70 MIDI Thru switcher (two in, eight out), and a programmable MIDI routing system (eight in, 16 out), the MM90.

Malcolm Cecil demonstrated the Electronic Music Publishing House's MIDIPlay program for the Atari ST. By keeping the program incredibly simple, Cecil has managed to make room for 200,000 MIDI events in MIDIPlay. Even though there are virtually no editing facilities, the low cost (under $50) and high note capacity make MIDIPlay an attractive prospect for performance situations.

Dr T's booth was busy as usual, and the company was showing their DX-Heaven and 4-Operator Deluxe program editors for the ST, as well as their new Mac Plus/512K Mac-based program, Views, which is unlike anything you're likely to have seen before. It's a 256-track MIDI 'tape deck" providing musical notation which includes all aspects of the recorded music, including MIDI parameters, pitch-bend, velocity and so on. Is this the Doctor's follow-up to the KCS event list method (see review elsewhere this issue) of sequence editing? The program also accepts Opcode-format sequence files, and operates with all Macintosh MIDI interfaces.

Southworth had their Jam Box up and running and chasing SMPTE timecode faster than anything we've seen so far. To prove several points at once, they had Jam Box working with Opcode's sequencer as well as with their own. For those who missed out on the announcement of Jam Box last year, the unit is a four-in, four-out MIDI interface for the Macintosh which allows simultaneous recording of four MIDI instruments, or four groups of instruments.

And speaking of Opcode, the newest additions to their lineup of Macintosh software include the Deluxe Music Construction Set (Version 2), program editor/librarians for the Yamaha FB01, Kawai K3, Oberheim Matrix 6/6R, and the Akai MPX820 mixer. The show was also scene for the debut of Opcode's Score! program for film and video sound production, which allows you to spot all your cue times, compute possible tempos, and lay out bars to fit in with specific hits. This will be updated in the Spring to include both SMPTE software (free) and hardware (additional charge), so that what you set up will be directly writeable to a SMPTE master unit.

Perhaps the most surprising of the lot were the various music programs on show for the Commodore 64. One might have thought that with all the more powerful 16-bit computers now available in the under-£1000 range, the C64 wouldn't have much appeal these days. But there are plenty of those computers still out there in the hands of musicians, so Steinberg have come out with the Card 32, a MIDI interface/sequencer/editor/scorewriter combination package, the SES DX editor, and the Cosmo Casio CZ editor. The Card 32 is of particular interest, since it features the most recent version of the Steinberg Pro 16+ sequencer which, in conjunction with the Scorewriter program also included, follows the music-creating process from the recording of the first track right through to a printout of the finished composition. Since these programs are held in ROM in the Card 32, there are no disks to be loaded except for those containing sequence files (important, when you consider how slow the Commodore 1541 disk drive can be).

And the Card 32 itself is probably the most versatile MIDI interface available for the Commodore 64, as it includes three MIDI outputs, an external clock sync and tape sync interface.

Steinberg are also continuing to develop new Atari ST software. On display were the Pro24 sequencer and MasterScore programs, accompanied by their new SMP24 SMPTE interface for the ST, featuring two MIDI inputs, four MIDI outputs, clock and SMPTE in and outs, a Centronics port, and a variety of sync, master, and MIDI programs and configurations.

Steinberg also have two Atari-based Soundworks sample editor programs for the Akai S900 and Ensoniq Mirage samplers, so no shortage of innovation there.

IN THE AREA of signal-processing, there was plenty of activity. Not least on the Roland stand where the company's Boss division had at last added a digital reverb to its Micro Rack series of compact effects. The RRV1O is a 12-bit, mono-in, stereo-out device whose available reverb effects include two rooms, two plates, multi-tapped delays and one gated mode. The maximum decay time, continuously variable, is ten seconds, and a built-in pre-equaliser allows tailoring of the reverberated sound.

Following in the footsteps of the Roland DEP5 comes the DEP3, which is basically a reverb unit that also offers delay and EQ facilities. The DEP3 is MIDI-compatible, has 99 memories and uses 16-bit signal processing to achieve a wide dynamic range with a 12kHz bandwidth. The maximum delay time is 450ms, and various reverberant environments can be called up including gated reverb, rooms, halls and plates. Other tantalising innovations from Roland included the VP70 vocal processor, an immensely versatile machine capable of harmonising, pitch-shifting effects and pitch-to-MIDI conversion, and the GP8 guitar processor mentioned above.

Alesis were showing the Microverb and the brand-new MIDIverb II, which incorporates the best ideas from the MIDIverb and MIDIfex, and adds chorus and flanging. Unlike the original MIDIverb, the new unit is rack-mounting, uses 16-bit electronics, and offers a 15kHz bandwidth. The 99 preset effects, assignable to any of 32 MIDI patch numbers, may be called up from the front panel or remotely via MIDI on any of 16 channels.

ART were showing their extensive range of digital reverberators, the most recent of which is the Pro Verb, capable of creating a wide range of preset reverb and delay effects. More radical, though, is their Smartcurve programmable graphic equaliser, which boasts 120 programmable memories and a user-friendly liquid crystal display. The equaliser is MIDI-compatible, and has a composite video output that allows you to see frequency response, equivalent slider positions and MIDI system information on any colour monitor screen.

Korg were also making a big showing in the signal-processing arena, with their DRV2000, a MIDI digital reverb with "multi-modulation". What this splendidly meaningless piece of jargonese refers to is the ability to route the 2000's reverb parameters to MIDI performance control data such as velocity and aftertouch, so that the louder you play your keyboard, the longer the reverb time (say) becomes.

Then there's the DRV3000, a multi-effects processor in the SPX90 mould which allows you to stack two of its treatments either in parallel or in series, and which can also be programmed using one of the slickest remote control units yet devised for a low-cost signal processor - handy for mounting atop mixing console while the DRV3000 itself is tucked away neatly inside a rack.

BUT two of the most fascinating developments of this year's Winter NAMM extravaganza made a bigger impact because they were unexpected - though for entirely different reasons.

The first of these came from US company Kahler, who, with all due respect, are not exactly known for their expertise in the high-technology arena. Yet, as occasionally happens when a new group of people addresses an area of interest from outside it, Kahler has succeeded in creating a machine which nobody else had considered marketing, yet which is exactly what many musicians and record producers want.

The machine in question is the Kahler Human Clock. Externally, it's an unassuming, rack-mounting black box; internally, it's simple enough - a converter system that takes any rhythmic pulse from a drummer, bass player or other musician, and through an exclusive process known as "Real Time Prediction", calculates the live tempo of the pulse and converts it into MIDI clock information. This MIDI output can alter instantly in tempo, and can, of course, be used to drive any MIDI drum machine or sequencing system.

The applications of the Human Clock are pretty extensive: you could use it live, with the band's drummer dictating the speed and "feel" of a MIDI sequence, instead of vice versa; or you could use it in the studio to replace a sync track that had been accidentally erased, or to "tap in" a sync track where one had never before existed.

The second fascinating development brings us back to the instrument that began this report - the polyphonic synthesiser. It's an entirely new digital synthesiser, with an incredibly low price-tag and a sound that is simply breathtaking.

It's called the Roland D50, and it was the one single instrument that had everybody, but everybody, talking about it.

The D50 uses what Roland call "Linear/Arithmetic Synthesis" to create its sound, though the precise workings of this system are not too clear at this stage. What you get on the D50 are two types of sound building blocks, known as "partials", with which to create your sound. The two types are PCM samples (which cover just the leading edge of the sound) and digitally-synthesised waveforms (for the rest of the sound).

There are over 100 of the former onboard the D50, and they include the attack portions of such things as bowed and plucked strings, samples of piano and mallet instruments taken from Roland's own SAS digital pianos, and the "chiff' of wind and brass instruments, as well as looped waveforms and noise. Each PCM component can be modified with its own five-stage digital envelope generator.

The digitally-synthesised partials can use sawtooth or variable pulse waveforms, and like the samples, have their own digital envelope generators for shaping.

You can use a joystick (similar to that on the Prophet VS) to combine up to four different partials of either type in real time, and send the result through a digital simulation of an analogue filter.

After that, your sound can be subjected to digital signal-processing by stereo chorus, parametric EQ and reverb, the settings of which are all programmable for each patch you set up.

Only after the processing stage (ie. at the audio output) does the sound of the D50 enter the analogue domain for the first time, so frequency response and dynamic range are of a high order - something that was confirmed by demonstrator Eric Persing's impromptu (he'd had the machine only for a matter of days) performances during and after NAMM.

It's possible (though not really constructive) to view the D50 as a compromise instrument that gets around the problems of most conventional methods of sound-creation by combining elements of all of them. But sometimes compromises are what work best, and although the D50 is too complex to have a completely foolproof programming system, the audible rewards for delving deep into its internal architecture are considerable.

RRP in the UK should be around £1450, with an "analogue" programming module (the PG1000 - lots of helpful knobs and sliders in the Roland tradition) available at extra cost.

Suddenly, compromise isn't a dirty word after all.