Dueling Oscillators: The making of Cursus and Loquelic Iteritas (Or: why everyone needs multiple oscillators)
We've done several demos over the past few months where people have asked us about how Cursus Iteritas and Loquelic Iteritas compare. As the designer, Stephen puzzles over this question, but as the other half (tester, user, general observer), to me, this seems like a pretty reasonable ask. In our demos, we focus on letting users put hands on the modules and see for themselves, but since not everyone gets to make it to an event, I'm going to put some of this together here.
Origins and Development: The backstory
Loquelic Iteritas (LI) took a pretty long time to develop. The original impetus for this module was just to put together something that sound designers and musicians could both use: it wasn't really meant to be a harmonic-sounding oscillator from the start. (You'll note that we succeeded on this goal.)
We started with a simple implementation based on an algorithm called VOSIM. VOSIM (VOcal SIMulation) is an algorithm designed in the 70s. It was meant to emulate human voice and models sound as squared sine waves. Together, these do a good job of mimicking the glottal pulses characteristic of human sounds...but being an early attempt at emulating human voice, it was far from perfect. For our purposes, we knew it would make some pretty cool sounds. But who wants a one-trick pony? At Noise Engineering, we prefer our ponies to know lots and lots of tricks. So we added two more modes: Summation Synthesis (SS) and Phase Modulation (PM).
All three modes were developed with a variable sample rate; aliased power is moved into frequencies that are related to the fundamental frequency. Despite this, the algorithms themselves are all fairly simple and have been intentionally left minimalistic: you can get harsh and interesting sounds from LI that we haven't really seen from other modules on the market.
Here's the basic demo of LI:
Cursus Iteritas (CI), on the other hand, was designed from the start with a different function in mind. Some time ago, Stephen was sitting around chatting with Scott Jaeger of the Harvestman and Yasi Pereira when they mentioned the Walsh functions. It was an interesting and weird math challenge and Stephen was taken in pretty fast, but the module got shelved for a while while other things took priority. When it came back up to the forefront, several things were pretty clear: first, that it would be an oscillator that generates wavetables on the fly; second, that it would be a far more harmonic module than LI; third, Walsh functions are complicated, with a large number of variables that needed to be reduced to something manageable. Some testing suggested a bandpass-like filter control could do the job. Finally, we knew that like LI, it would need to be a pony of many tricks.
The next step was to find a couple of other modes that could generate sounds in the context of the bandpass controls we developed for the Walsh transforms. Fourier was a no-brainer: it is based on sine waves and mapped easily to the controls. Finally, Stephen turned to more recent mathematics and the Daubechies 4 wavelet function. The resulting waveforms are somewhat sawtooth-like and create musically interesting sounds.
Here's a basic demo of CI:
The manuals for each module have far more detail, but if you still have burning questions about the differences between LI and CI, please contact us. If you have other ideas for posts you'd like to see here, please also drop us a line!