SAMPLING IS DEAD, long live sampling. When the history books are written, I'll wager that the arrival of resynthesis will mark the beginning of the end for our friend, the sampler. What's more, I really don't think it'll take that long. Witness the Acxel.

What exactly is resynthesis? Resynthesis is the process of analysing a sound and arranging a number of sound components, like oscillators and amplifiers, to reproduce that sound. In other words, a sound is digitally sampled and then analysed and recreated with a series of sine wave oscillators (as in additive synthesis) whose pitch and frequency change over time. If you have a sophisticated additive synthesiser, you could create the sounds manually, but the beauty of resynthesis is that this process is automated. It selects the basic frequency and amplitude of the various harmonics (the number of which varies according to how many you have available, though the default amount is 32 per voice on the Acxel) and creates envelopes which vary these levels over time.

The advantage of resynthesis over sampling is that while samplers treat sounds as a whole, resynthesis involves breaking a sound down into each of its component parts, or harmonics. The components are all independent and can be adjusted individually, so you have more control over the sound. Ideally, a resynthesiser will sound as good as a sampler, but it will provide additional control over the sound.

Ever tried to copy the violin scrapes and breathy attach of the D50? If so you'll have found you can't do it using simple sampling - you still hear the fundamental pitch underneath the effect. With resynthesis you can remove the pitched component of the sound (ie. the fundamental and perhaps a few of the other first harmonics).

Another problem overcome by resynthesis has to do with the link between pitch and duration of samples.

Transposing a sampled sound upwards by one octave cuts the length of the sample in half and transposing it downwards by that amount doubles it. With resynthesis pitch is independent of duration, consequently sound duration may remain constant over the entire range of a keyboard, and time expansion and compression without pitch change become possible.

And this is what the Acxel is all about. The name Acxel, by the way, stands for Acoustic Element. In the company's terminology, each harmonic of a sound (represented by an individual sine wave) is referred to as an Acxel. The individual Acxels are then "added" together to form a complete, resynthesised whole. The Acxel is also an additive synthesiser and if you don't buy the optional Acxelizer, which performs the resynthesis process, then you're left with a very extensive additive synthesiser.

Back to nuts and bolts. The Acxel consists of two basic parts: the Solitary, a large rack-mounted black box which holds the system's plug-in cards, a 3½" floppy drive, 2Meg of RAM and the various audio, SMPTE, MIDI, CRT terminal, printer and hard disk connectors; and the Grapher, the company's revolutionary interface. Future plans also include an optional high-quality dedicated keyboard controller.

The Grapher is an impressive, touch-sensitive terminal that features an 80-character LCD as well as a 32X64 matrix of red LEDs which can display waveforms, envelopes, relative harmonic levels and a host of other functions. All of the "switches" on the Grapher which surround the rectangular matrix and all of the points on the matrix itself, turn on or off with the touch of a finger, making the system very fast - drawing an envelope or a waveform is simple and extremely intuitive. Once experienced you can't help but believe it's ideal for working with sound data. It's similar to a light pen or a graphics tablet, but more sensual. It's also ideally suited to the Acxel because of the speed with which you can adjust a great deal of data. Best of all, the Acxel responds to adjustments to the sound in real time and, consequently, the Grapher can be used as a performance tool. Significant real-time sound shaping has finally arrived. (The actual resynthesis calculations take a bit longer; roughly two seconds for a two-second sample.)

The voice architecture of the Acxel is rather complex, but it does retain ties with traditional synthesiser voices. The basic sound components are created by Intelligent Synthesis Cells. Each ISC consists of an Intelligent Digital Oscillator (IDO) and its accompanying Intelligent Pitch Envelope Generator (IPEG), as well as an Intelligent Digital Amplifier (IDA) and its accompanying Intelligent Volume Envelope Generator (IVEG). The reason the word "intelligent" precedes all these components is because they respond to the data generated by the Acxelizer from a sample and intelligently program themselves to appropriate settings. A normal voice consists of 32 of these ISCs, but the number can be reduced to 16 or raised to 256 for varying degrees of harmonic resolution. The total number available in a system ranges from 128 to 1024.

In typical additive synthesis or resynthesis the oscillator would produce a sine wave, but each IDO on the Acxel can produce a completely independent, user-programmed waveshape. Consequently, you should be able to produce complex sounds with relatively few ISCs as well as be able to create noise with a single oscillator. The oscillators can also be independently detuned for chorusing, which the company refers to as the Harmonic Rainbow effect. The envelopes are also a little more sophisticated than usual; they can be independently delayed, and adjusted by a number of real-time functions. Each envelope offers up to 1024 steps.

But that's not all. After the sounds of the ISCs have been added or mixed together, the composite signal can be affected by two Digital LFOs (DLFOs), one of which is Intelligent, an Intelligent Digital Filter (IDF), an FM processor with an index envelope, a Master IDA and an analogue low-pass filter. The IDF can function as any type of filter including a Variable Integer Pass (VIP) Filter, which only affects certain harmonics of the sound. It can also be programmed and adjusted by the user in real time either with the Grapher or with other controllers.

In addition to traditional waveforms, each of the LFOs can have a user-defined waveform, and each has its own dedicated amplifier with a multi-stage envelope. DLFO1 is also intelligent, which means that it responds to the resynthesis process to recreate a vibrato effect or, if an entire musical line with different pitches is the sound source, it recognises and recreates the changes in pitch.

The Acxel isn't cheap, but considering that the only other commercially available device offering resynthesis is the Synclavier, that's not surprising. The system currently being produced (called the Pro Studio) features eight modules of 32 ISCs each, the Acxelizer, and eight individual outputs (unfortunately, stereo outputs aren't available). You can get into the system for little more than half the price of that, but you won't get the Acxelizer, you'll only have four modules and there's only a single XLR output.

The Acxel sounds impressive on paper, but the bottom line is how it sounds, and that still remains to be seen. Is the resynthesis process accurate? Does it sound as good as it promises? Stay tuned for the answers.

Prices starter stage (without Acxelizer), £8500; starter studio. £14,545; Top of range system. £37,930

(Contact Details)