SAMPLING IS NOT as easy as it looks; something a large bundle of us are about to find out. Akai's laudable success at bringing in a polyphonic sampler at the above price will make the technique available to many more musicians. But if you thought you could point a microphone at the London Philharmonic and have a top ten single three weeks later, then the S612 will also be delivering certain rude awakenings.

For those unacquainted with the Trevor Horn phenomenon, a sampler digitally records an 'acoustic' sound so it can be played back at a different pitch from a keyboard. Some digital delay lines can do that but a true sampler (the Fairlight and Emulator being the grandaddies) can produce polyphonic playbacks from the single snatched sound.

But creating a great sample — in particular, a great sustaining sample, not just a percussive one shot — is no quick fire act. You need to learn what sorts of sounds will successfully loop back on themselves to produce a steady tone for as long as your fingers are on the keys; a tone unscarred by glitches, hiccups or clicks.

And that's where the education will be going down, brother. It's a safe bet that many synth manufacturers' R&D departments are presently working on editable sampling systems — those where you can get at the digitally recorded data and alter the sound from the inside. But the Akai is what we have now, it's well priced and it's a fine job.

Unlike the American Ensoniq Mirage — the Akai's Frankfurt rival — the S612 doesn't have its own keyboard but links MIDI-wise (in, out and through), to a synth of your own. It's six note polyphonic but can be converted to mono by one of the 11 membrane switches on the front panel. The MIDI formating and appearance make me suspect it's aimed at Akai keyboard owners (obviously) and Yamaha men. It's velocity sensitive and dovetailed perfectly with the office DX7.

Sampling is 12 bit, and basic — none of the keyboard splitting, memories, or different sounds for different notes facilities of Emulators and Ensoniqs. Once your noise is recorded (via mike or line jacks sockets at the front) you can trim off the sample's beginning or end by using two start and finish sliders; isolate a section in the middle or reverse the playback by the same method.

These sliders are essential. When you're inputting your own 'sample' — a slapped bass string, a drum thwack — you may get away with one clean take. If you're ripping off... sorry... extrapolating from an album or otherwise previously-recorded track you'll invariably want to slice off spare notes at the beginning and end, or tighten up the attack.

Running the Akai from a MIDI sequencer (say, during a Portastudio session), leaves you free to shift the sliders during an, ahem, 'performance'. This will drag many varied sounds from one sample often by shifting the sliders a mere fraction.

The Akai offers three ways of playing back your recording. As a 'one shot', it pumps out a single complete playback of the recording at a pitch (or pitches) dictated by the connected keyboard. As with all such devices, it achieves this melodic reinterpretation by speeding up or slowing down the replay. The irrevocable side effect is that a one-second-long sample at middle C will take two seconds to play out one octave down and half a second to play out one octave up. (You very quickly learn that widely spread chords end somewhat unevenly.)

Version two is looping — the perpetual repetition of the closing stage of a sample for as long as your fingers are on the keys. There's no point repeating the whole sample — you wouldn't want to glue the slow attack of a violin note back on the end of its crescendo.

Here's where your choice of sample (and the slider editing) comes in. If the concluding moments of your recorded sound pass through significant changes in dynamics or tone, the loop will hiccup. If those moments are reasonably static, you stand a better chance and you home in on those moments with the sliders. This was the toughest struggle with the Akai — not surprising since it's the greatest problem area with any sampling system. An 'unlucky' loop can degenerate into a flat whine not unlike feedback. Success (for me, anyway) depended on a lot of juggling with the sliders, and several goes at the sound to be sampled.

The third choice is 'Alternative' which grabs a smaller slice of the recording and loops backwards and forwards with it — sometimes better, sometimes not, but invariably with an undulating effect to it.

Speaking of which, one way to conceal a loop join is to add vibrato. On chords the 'splices' are all coming around at different times (each note plays back at its own speed, remember) so any lurch in sound is already masked. The Akai's LFO section — speed, depth and delay — also produces faster and slower vibrato depending on the note held down. Usually there are enough random wobblings to water down the glitches. Next time you cop a choral or orchestral sound on record with an indefinable meandering to it, you'll know why.

Faced with the posers inherent in sampling, you soon begin to recognise other people's escape routes. Take that habit of playing one note, staccato, half-a-dozen times so you hear only the first instant of the sample. It's irresistible.

And my most successful experiments were in sequencing the Akai. There's a magnetic attraction to writing a part and loading in various samples, then tweaking them with the sliders until you get a completely unexpected (and magnificent) result. Thank you to Phil Collins' first tom fill from 'In The Air', the bells/bass drum in Depeche Mode's 'People Are People', (even better when it was reversed by putting the start slider past the end slider) and the first few bars of Mars from the Planet Suite.

And you only need one decent note out of your decrepit bass guitar (Di-ed through the line socket or with some ambience if you want), and you're a long way towards 'Two Tribes'. Recording levels for mike and line are set up via a seven segment LED meter with controls for input and monitor. The final, three red segments do mean danger — distortion can result.

Lifting material from an album or tape involves fixing the level on a run through and punching the record button as the desired section comes up. The Akai will trade off sampling time against frequency response. Sampling rate is 32kHz with 12kHz as the cutoff frequency. The longer the sample, the less treble makes it home, and at the full eight seconds, the world is very dull indeed. Pre-selecting the recording time is neatly done by pressing a note on the keyboard before sampling starts. The lower the note, the longer the sample.

Inputting your own sound is simpler, but can have traps. Press record, and the Akai will wait, quiescent, only turning itself on when it hears your drum bash, guitar twang, etc. If you're working behind the Akai, a blinking red light on the back panel tells you it's armed. Be careful of sounds with slow attacks as the Akai might miss the all-important moment of origin. Safest to make a quick, loud noise, wait a split second, then play your piece. You can edit out the noise and the gap later.

For the closing touches there's a filter control to brighten the output (best left on full unless you need to silence a violently hissy sample), and decay. The latter will give your looped sounds a slight artificial release time once you take your fingers off the keys. Stops the sound ending dead — a well thought out addition, and invaluable.

The Akai receives on only 9 MIDI channels, and can dump successfully made samples onto tape. There's a third socket for an external trigger and the playback can be master tuned a semi-tone either way.

The last facility is overdub — piling one sample on another. Theoretically, you could go on forever. Practically, one or two extra passes may thicken a sound (eg: two goes at the same synth patch, but a slight change in tuning second time round) or combine wildly opposed elements but more than that is not always advantageous. It's not possible to 'splice' two sounds back to back, just overlay them.

While you can compile ludicrous noises, there is a noticeable increase in hiss with successive dubs, and I still found the best results were lifting from cassettes or albums where sounds were already mixed. Too much aural activity quickly becomes confusing if you're trying to use it within a track. Listen to the radio and see how many times complex sampled sounds are completely isolated from the rest of the mix.

Finally, the sound quality. There have been those (whose acquaintance with circuit diagrams is greater than ours) who've described the Akai as a polyphonic version of the Electro Harmonix Super Replay (One Two, July issue, £500). Similarities exist, the Akai is quieter, definitely, but not to the degree of hiss-less-ness that would put it alongside plain digital delay lines in the mid-hundred bracket. Much depends on the care with which you make your sample, but a professional studio would raise an eyebrow at the noise figures.

The Ensoniq, when it comes, will trample over the Akai in terms of facilities — but it will be roughly twice the price.

In the time being the S612 is here, now. It works, and works well, if not to the highest pro-application specs, and Akai have come from the back to give the world what it wants — polyphonic sampling at a bearable price. As for the wisdom of the whole sampling principle in music... anyone looking for a new toy will be bored with the Akai in a day, anyone with some imagination will have a use for it as long as there's a use for sound.

A cast of thousands at a cost of one.

AKAI polyphonic sampler: £1099

CONTACT: Akai UK, (Contact Details).