proteo
shapeshifting stereo oscillator
The name, indeed, comes from ancient Greek mythology since Proteo was a shapeshifter who could assume whatever shapes he preferred. In this way, our Proteo, has not a precise sound on its own but a clever mechanism that permits to acquire and transform every signal into waveforms. What if you can use your envelopes, LFOs or any other signal as a sound source?Dimensions: 12hp
Power consumption:
+12 rail: 90 mA
-12 rail: 43 mA
+ 5 rail: 0 mA
What sets it apart from other oscillators?
What's very different in Proteo is its internal pseudo-tape recorder, which is capable of continuously acquiring and transforming any external signal into waveforms and morphing them from a basic sine.
In Proteo, waveforms are not a static digital representation of a sound; they can continuously evolve in a very organic way, and tape loop time is a fundamental factor:
changing it will drastically affect the resulting output even with the same sources at the inputs.
Slowing down the time will capture a bigger portion of the incoming source with more harmonics.
As a result, increasing it allows to capture up to low audio rate signals or really small fragments of the signal.
Exploring the interaction between Proteo, the sources and the time will force you to change your mind about how sounds and harmonics can be generated:
visualizing how different shapes can sound and looking for new sources to generate YOUR waveform is the fundamental idea behind Proteo.
Proteo is also stereo.
Thanks to the Span knob you can easily detune one oscillator from the other to create nice stereo effects.
Beside the common pitch the two oscillators have independent acquisition paths and morph controls, each one can emit different waveform.
how does it work?
By default, Proteo emits two sine waves at its outputs[8].
The left section is composed of a time-based acquiring path with two separate inputs[3],
insert here any source you want to acquire (CV, gate, or low audio).
Each input has its preamp[5], which can attenuate or double the amplitude of the incoming signal.
Acquisition time[6] is the speed of the digital tape recorder, from 2.5 seconds to 1kHz.
An external input[2] with a dedicated attenuverter[4] will let you modulate it from outside.
Morph[14][11], controls the interpolation between the sine waves and acquired waveforms.
The final waveform can be seen on the two small oscilloscopes[7], one for each oscillator.
Acquired waveforms can also be locked independently with the two freeze buttons[15] or with an external signal through the freeze gate ins[1].
The two oscillators are tied together to the same V/oct[9] and pitch, which can be changed with the freq knob[13].
Span[10][12] controls the distances between the left and right parts by detuning the right one from 4 octaves below to 1 one octave higher.
how we got there?
…some thoughts from our blog We started working on Proteo almost four years ago after discovering Daphne Oram's incredible Oramics, or, at least, what we thought it was.
To be honest, we initially misunderstood its mechanism. We could also say that Proteo is the child of some translating error, which we think is a cool thing on its own anyway.
Indeed, we initially thought she was drawing the actual waveforms on some celluloid film but, in reality, she was "just" modulating some oscillators by exposing several photocells with different "shading patterns" (which is still incredible for 1957)
We have always loved the idea of a synthesizer where you can draw waveforms, so we dropped the Oramics thing and went our own way.
The concept was a shapeshifter oscillator that, by default, has no sound on its own and creates waveforms by acquiring external signals in real-time from its preamps.
Initially, we designed Thererec and Proteo as a single module, with the drawing and the sampling parts in the same module.
Then, we realized splitting it into two dedicated units would be even cooler, so we created the two modules.
Anyway, we didn't want to assimilate Proteo into a sampler or a wavetable oscillator. We didn't want something where you create your own palette of sounds, save it in a memory slot, and recall it as a preset.
What we liked most was exploring the concept of mangling a waveform inside the time domain and threatening it like treating a micro loop on a reel tape.
You can capture different levels of detail by adjusting the recording speed.
In Proteo, this is the basic mechanism that forms the "wavetable".
However, while this is a very laborious process in an analog domain, inside the digital domain, this could be done seamlessly and continuously. And that's where the cool things start to happen!
As we quickly discovered, this process made it quite easy to create many nasty sounds with a huge load of evolving harmonics.
We decided then to add the MORPH control, which can blend the captured waveform with a sinewave.
In this way, the fundamental harmonic can still be heard trough.
Going a bit nerdy on the tech stuff, instead, we would like to say something on the engine we had to develop for it.
As big fans of analog, old-school oscillators, we have gone through many trials to ensure that Proteo matches that sound quality.
With the classic fixed sample rate architecture, much of the captured harmonic content was lost due to spectrum limits.
A lot of filtering was needed to avoid artifacts like aliasing, and, in the end, it was always dull and lifeless.
We didn't like how it sounded so we focused more on other projects.
Speaking about “drawing waveform”, in the following years we also considered the idea of making something “easier” in the old-fashioned analog way:
something like a reproduction of the mighty Buchla 132.
This idea did excite us, but in the end, we wanted the digital domain's flexibility to give it some new functions and the ability to actually draw waves.
For a long period, an early prototype of PROTEO lay around the lab unused.
Things changed when we tried another digital “reproduction” method (which is quite obsolete in 2023) called "variable sample rate".
With this method, you're not sampling a variable quantity of points inside of a certain wavetable, but you are actually playing all of them at different speeds by changing the clock rate of the internal microprocessor.
This is much closer to what happens when mangling a loop in the analog domain. Most importantly, there's virtually the same resolution at every speed, with no loss of details on the high spectrum and no aliasing at all!
Finally, we had the sound quality we were aiming for!
As a side note, we didn't invent this, but this is the same process used on early wavetable synthesizers like PPG WAVE, synclavier, and EMU, just to name a few. It is also very similar to how old EEPROM-based drum machines worked! (Linn Drum, Obhereim DMX, Roland 909...)