The TECHNOS ACXEL Resynthesizer

Recently, we took delivery of a very rare and unique synth whose perspective on sound creating would be considered unique. The Technos Acxel was created in the late 1980s and holds a strong position in the hierarchy of digital synthesis. We asked Ryan Gaston of Perfect Circuit to give us an analysis of this unique machine.


The Technos Acxel is, with little doubt, one of the most peculiar instruments I have ever encountered. Living in 2019, it feels safe to say that an instrument with its combination of qualities would be highly unlikely to emerge in this day and age. It focuses on complex and obscure synthesis methods and has a strikingly peculiar design: a design which, while quite well-suited to its rich (albeit quirky) internal architecture, would no doubt have been quite expensive to develop/produce…and thus, quite expensive for the end user. Technos was clearly driven and ambitious, sparing no expense to pursue their vision of what making music could be like—and this ambition led them to take significant risks.

The Acxel was one such risk, and one that sadly did not end with fame or fortune: approximately 35 of these instruments were produced before the company closed its doors in 1992. The name “Technos” faded into obscurity and, aside from occasional mention in historical texts (such as Mark Vail’s Vintage Synthesizers), garnered little or no continued attention. To my knowledge, the Acxel as of yet has produced no Minimoog- or OB-Xa-style chart-topping hits; it didn’t appear on stage among elaborate prog rock or synth pop groups; it didn’t find a home in any giant, acclaimed Hollywood studios. Instead, it found its way into the homes of experimental synthesists and/or collectors, and few seem to have publicly surfaced since.

________

Personal History with the Acxel

I was working at Perfect Circuit in Burbank, CA when I first encountered the Acxel. At that exact time, we had a host of odd and uncommon synths: a Yamaha CS-80, a Synton Fenix 1, a Waldorf Wave…all exciting things for any synth enthusiast. Among the pile of things on our testing bench, though, was an Acxel. I had never seen or heard of it before, and at first glance was not even sure that it was a musical instrument. I was tasked with learning how it worked, so as to be able to determine whether or not it was actually in functional condition.

After several days poring through manuals and testing features one by one, I gained a profound respect for the instrument. I maintain that I have yet to encounter an instrument with such a singular workflow—evidence that its creators at Technos had a very complete and sophisticated way of conceptualizing sound. It is completely distinct from the then-mainstream approaches of subtractive synthesis, sampling, and digital methods such as FM or PD, instead borrowing bits of each approach to create something unique. It quickly became one of my favorite instruments, both due to its bold vision of how sound can work, and due to its clever interface—but more on that soon.

That instrument eventually found a home with a private collector, and I accepted the fact that I would likely never see one again. But, two years later, another found its way to our shop. Happily, this instrument eventually wound up with EMEAPP: a perfect fate for something so peculiar. All too often, rare instruments such as this circulate within closed networks of collectors and eventually fade from public visibility altogether, resulting in an unintentional erasure of the fringe areas of historical instrument design. And while these boundary-pushing instruments may not be the most influential when considered in terms of widespread use, they often have powerful, unique potential to challenge the way that we think about instrument design, sound design, and music-making altogether.

________

The Technos Acxel Itself

The Acxel is comprised of two pieces. The Solitary, a large, rack-mountable chassis, houses the instrument’s signal processing and audio/MIDI I/O boards. In practice, the user does not interact much with the Solitary, aside from loading the OS or timbres from a 3.5″ floppy disk. Perhaps comically large by modern standards, the Solitary houses an astounding array of electronics: including basic I/O boards, the Syncards (voice cards), and depending on the exact model/configuration, the Acxelizer board itself.

The second piece, the Grapher, is the actual user interface…which is to credit for most of the Acxel’s cult reputation. The Grapher is a tablet-like control surface that allows simultaneous display and editing of all of the Acxel’s internal parameters via a series of over 2000 small LED-embedded capacitive touch plates. The Grapher’s Matrix of touch plates changes in function depending on what parameters have been selected for editing: so the Grapher is at once a menu/page-based display and a means of changing values at the touch of a finger. One can think of this as being somewhat like the Fairlight’s light pen—but it is activated simply by touch. Again, this is the way that all internal parameters are edited: envelopes, wave shapes, signal levels, and much more are all created and displayed through this absolutely sci-fi-level touch interface.

The Acxel does not provide a dedicated playing interface, though: it must be paired with a MIDI controller. This was a departure from Technos’s prior design, the 16pi, which was itself not dissimilar to an Acxel with a keyboard. Instead, users were to determine their own means of interacting with the instrument.

_______

What is Additive Synthesis?

But the user interface is not the only peculiar thing about the Acxel: its means of sound production is also markedly uncommon. The 1980s were a bold time in instrument design; designers’ increasing reliance on digital technology for sound generation and control meant that, theoretically, almost any musical concept could be turned into an instrument. As such, some designers turned to theoretical techniques that previously might have seemed impractical—including additive synthesis.

Mainstream analog synthesizers primarily rely on subtractive synthesis—that is to say, they rely on harmonically rich oscillators (saw waves, square waves, etc.) as a sound source, then using filters to remove harmonics and emphasize certain parts of the sound spectrum in order to create interesting sounds. This is the realm of the Minimoog, ARP 2600, Oberheim SEM, and countless other classic analog synths. Additive synthesis, however, works differently: rather than using a small number of harmonically rich oscillators and a filter as building blocks, it instead typically uses a large array of sine oscillators carefully tuned to specific intervals and set to specific volumes in order to construct a sound from the ground up. This is a somewhat more atomic/elemental approach; and because a convincing additive “voice” might require dozens of oscillators and amplifiers,  additive synthesis was too impractical to implement and too tedious to program in an analog context. However, when designers began to rely on digital sound generation and digital control, such a feat was suddenly within reach.

This is the fundamental approach of the Acxel. This LED-laden oddity has a whopping 32 oscillators per voice, each with their own amplifiers. Each voice is rounded out with two-operator FM capability, two filters, and a master amplifier for voice-level volume control. Moreover, there are 64-stage envelope generators for practically every sensible destination—each oscillator in each voice has its own pitch and volume envelopes, and several voice-level envelopes control FM index, filter cutoff frequency, and overall voice amplitude. Yes—there are literally dozens of envelopes for every single voice (and the basic configuration has eight voices in total).

Of course, having this sheer number of sonic resources in one instrument is pretty jaw-dropping…but one of the most fascinating aspects of the Acxel lies in how all of these resources are controlled.

________

Additive Synthesis in the Acxel: the Spectrum Concept

Technos cleverly organizes the key parameters of additive synthesis into spectra in order to be able to understand the relationship between settings for each of a voice’s internal oscillators. There are several spectra: the amplitude spectrum, phase spectrum, integer spectrum, and beating spectrum. When navigating to any of the pages for editing these spectra, the touch plate LEDs illuminate in order to reflect the current settings for each oscillator in order, basically creating a touchable bar graph representing all of the settings for the selected spectrum parameter. One needs only to swipe their fingers across the Grapher’s touch plate matrix in order to alter these settings, as if it were a set of sliders or drawbars.

The amplitude spectrum defines the relative loudness of each oscillator, allowing users to emphasize or de-emphasize any overtone they desire. The phase spectrum determines the phase offset between each oscillator: the effect of the phase spectrum is more pronounced when several oscillators are tuned to the same pitch, so the effect of the phase spectrum is generally quite subtle in more complex sounds.

The integer and beating spectra work together to define the pitch of each oscillator. The integer spectrum can be thought of as a more coarse tuning value, with the beating spectrum providing finer control. The integer spectrum multiplies the base MIDI pitch by integer values from 1–32 as a means of easily tuning oscillators to precise harmonic ratios; this makes it easy to create harmonically pure, stable sounds. The beating spectrum, on the other hand, is a detuning control: it allows users to offset any oscillators’ pitches from their current “integer” values in order to access all of the inharmonic pitches between ideal overtones. This allows for much more complex, nuanced, evolving sounds…and even using these four spectra alone, a wide range of peculiar and novel sounds is available.

_________

Dynamic Additive Synthesis

But there’s more to it than that—as I mentioned before, the Acxel is overridden with envelopes. And these aren’t your garden-variety ADSRs: they are 64-stage envelopes, each with user-definable hold and loop regions. As mentioned above, there are voice-level envelopes for controlling filter cutoff, FM index, amplitude, etc…but there are also per-oscillator envelopes specifically for controlling pitch and amplitude. Yes—there are roughly 70 envelopes per voice. Ordinarily, handling 70 envelopes would be an overwhelming and tedious feat; but as with the clever organization of spectra, Technos came up with a relatively efficient (and dare I say, almost intuitive) way of handling envelope settings.

Amplitude and pitch control in the Acxel can operate in one of two high-level modes: Fixed Additive Synthesis (FAS) and Dynamic Additive Synthesis (DAS). In FAS mode, amplitude and pitch values are determined by the spectrum values alone, but in DAS mode, the envelopes come into play. Users can switch additive synthesis modes and access envelope editing pages using the ISC (Intelligent Synthesis Cell) menu touch plates on the left of the Grapher’s touch plate matrix.

Envelopes are assigned to individual ISCs (the combination of one oscillator and amplifier) by using the Numeraline, the row of 32 touch plates immediately below the Grapher’s matrix. Once a cell has been selected for editing, its current envelope shape is displayed via the matrix LEDs: it can be re-shaped simply by swiping one’s fingers across the matrix surface. What is particularly interesting is that multiple ISCs can be selected for editing all at once—by selecting multiple ISCs via the Numeraline, it becomes easy to make broad changes in several cells at the same time. This removes much of the tedium in different implementations of additive synthesis, and makes it fairly quick to achieve complex, dynamic results.

DAS control of amplitude is fairly straightforward: it allows evolving changes in the concentration of particular overtones, providing easy access to continuously shifting spectral focus. There are two separate DAS modes related to pitch, only one of which may be used at a time: DAS-I, in which envelopes modulate the current Integer Spectrum values, and DAS-P, in which envelopes modulate the current Beating Spectrum values. DAS-I produces considerably more coarse pitch changes locked to harmonic ratios, often resulting in peculiar harmonic arpeggio-like effects. DAS-P, on the other hand, is an excellent way of introducing dynamic detuning, twisting overtones into constantly changing inharmonic ratios for sounds that shift from stable to clangorous and back again. By using intentionally-devised combinations of pitch and amplitude envelopes, the Acxel can produce remarkable variation across the course of a single note. (As a side note: in my personal experience, the Acxel excels at drawn-out, atmospheric/textural sounds…there’s simply so much possible within a single note that can too easily go unnoticed with shorter, more idiomatic “keyboard” sounds.)

As mentioned before, there are also per-voice envelopes for volume, filter cutoff, and FM index. All envelopes in the Acxel are 64 steps long, can range from hundredths of a second to half a minute, and allow the user to define a region of segments to loop when a key is held (using the Alphaline, a line of touch plates above the Matrix). What’s more, all envelopes utilize what Technos calls the Dual Envelope Concept: meaning that each envelope can continuously interpolate between two different, user-definable shapes. By using key velocity to perform this interpolation, for instance, the Acxel can be made to produce highly dynamic results which, in the best situations, can lead to combinations of character entirely impossible to predict.

________

The Tandem Problem & Resynthesis

The use of Spectra and the Acxel’s clever interface for editing envelopes does make for an uncommonly smooth and immediate additive synthesis experience—but it’s no secret that even this implementation is far from the ease of use of a Minimoog, for instance. Making good sounds from scratch can still take a considerable amount of time, cleverness notwithstanding.

Happily, the Acxel does offer another way of defining sounds: resynthesis. In many ways, resynthesis is the Acxel’s primary intended use case—or at least, that is what sales literature would leave one to believe. Technos saw this technique as a solution to some of the issues inherent in the then-still-maturing technique of sampling: namely, the issue they describe as The Tandem Problem.

The Tandem Problem is no doubt familiar to anyone who has used a traditional sampler. In samplers, pitch changes are often achieved by using playback rates different than the audio’s sampling rate. By playing back at rates slower than the sample rate, output sounds appear lower than the original sampled sound; by playing back at faster rates, the sound appears higher in pitch. This is great (and useful in a lot of situations!), but comes with an intrinsic, unintentional consequence: it also results in the sound being shorter as it increases in pitch, and longer as it decreases in pitch. This is the Tandem Problem.

Resynthesis is related to sampling, but a bit more complex in execution. In resynthesis, a source sound is recorded into a buffer and then analyzed (often via FFT or similar methods). The result of this analysis is a mass of information about the sound’s harmonic structure and how the loudness and pitch of its overtones change over time. The resynthesizer’s internal synthesis engine, usually comprised of a huge number of sine generators and amplifiers, interprets this analysis and utilizes it as control information, resynthesizing the sound through additive synthesis. The end result is a sound ideally quite similar to the input analyzed sound.

This has a huge number of advantages over sampling. The first is that the Tandem Problem is no longer a concern: since sounds are constructed out of banks of variable-frequency sine waves with pitch and volume contours defined by variable-duration envelopes, changing pitch doesn’t have to mean changing the duration of a sound (and vice versa). Changing pitch is instead just a matter of shifting all of the overtones by an identical musical interval (maintaining their harmonic relationship). Similarly, changing duration is just a matter of expanding or compressing all of the envelopes within a sound…pitch doesn’t have to change at all. This means that pitch and duration changes are both possible, and entirely independent of one another.

________

The Acxelizer

The Acxel achieves this process via the Acxelizer: a processor based on a simple AI that performs the analysis described above and assigns all of the most pertinent information within the analyzed sound to the voices’ Intelligent Synthesis Cells—ISCs, the combination of an Intelligent Digital Oscillator (IDO), Intelligent Digital Amplifier (IDA), Intelligent Pitch Envelope Generator (IPEG), and Intelligent Volume Envelope Generator (IVEG). The Acxelizer can even determine how many overtones are required to adequately reproduce a sound and “steal” resources from multiple voices to construct a single sound…which ultimately reduces polyphony, but enables the resynthesis of considerably more complex spectra, even including highly effective resynthesis of the human voice. Another interesting implication of resynthesis in this fashion is that it permits extensive manipulation of the sound, down to a much more atomic level than with pure sampling.

The Acxel maintains the same level of editing potential with resynthesized sounds as it does with purely synthesized sounds…so users can reach within a sound and modify the spectral profile, and the way it evolves over time. This is excellent for creating alien sounds from familiar sources, allowing the cadence of natural tones to influence the unfolding of otherwise impossible textures. You can quantize the natural inharmonicities in a sound to harmonic intervals; you can frequency modulate resynthesized sounds; you can accentuate weak overtones and de-emphasize strong ones. Being able to reach down to this level of detail in recorded sound is all but impossible…and this is what is meant by “Acxel”: it is an auditory pixel, the smallest unit of digital sound to which Technos provides you access.

The Acxel came with several 3.5″ floppy disks of presets: a host of synthesized and resynthesized sounds ranging from Moog-style bass to harp glissandi and loon calls. Some of these sounds are far more compelling/interesting than others…and while they are remarkable given the age of the instrument, some have “held up” better than others. Regardless, the instrument’s capabilities to synthesize complex sounds from the ground up are remarkable…and only otherwise possible at the time through similarly complex/valuable instruments such as the Synclavier.

Note: despite the name, not every Acxel necessarily contains an Acxelizer: the most affordable model at the time of production was the Acxel Stage, which could synthesize sounds from scratch or use preset “Acxelized” sounds created with another Acxel. British magazine Music Technology reported in their March 1988 issue that the Acxel Stage cost 8500GBP at that time. The Starter Studio, a basic system with Acxelizer, cost 14,545GBP; and the “top of range” system, complete with Acxelizer and maximum number of voices ran for 37,930GBP. EMEAPP’s Acxel is what, according to sales literature, was called the Pro Studio model, containing an Acxelizer and one Syncard (for a maximum of eight voice polyphony), with both a mixed signal output and individual outputs for each voice.

________

There is so much more to the Acxel. It is multitimbral—individual voices can be split, layered, or operate on completely separate MIDI channels. It has an analog lowpass filter for each voice, and a Variable Integer Pass (VIP) digital filter which allows you to hone in on specific Integers; there are LFOs (including an Intelligent LFO for translating vibrato in resynthesized sounds); you can even change the wave shape of the oscillators used for resynthesis…the list goes on.

But after getting to know the Acxel, it becomes apparent that it was never quite finished. Several front-panel controls do nothing; several sections of the user manual refer to “reserved,” incomplete functions; redundant editing pages make it feel as if features were planned, but never quite made it into existence. It is clear that Technos had big plans for where the Acxel was heading…and despite all they accomplished, it doesn’t seem as if it ever got as far as they might have liked. Later iterations were planned, but sadly none ever came to fruition.

Despite this, working with the Acxel is an amazing and rare treat…one that I hope I’ll have again in the future. It is a true rarity, and an ideal packaging of what was great about instrument design in the 1980s: designers were bold, and truly sought to create instruments unlike anything else that had ever existed. And, in this case more so than many, they succeeded.