Amphibian audio, please.
Creating the foundation of a synthesizer must, in my mind, begin at its core: the oscillator or oscillators. This was the logic I started with when I began researching one of my favorite sound generating devices of all time, the SID chip used in the Commodore 64.
Luckily, the are quite a number of emulations available that have source code freely available, each one focusing on varying degrees of authenticity in their emulation of the original chip's circuitry, and from reading the source code from the various projects, and what I could decode from its circuit diagrams, I could find what made the original tick, so to speak, and I created a simple synthesizer able to recreate its sound. Now, this in itself is not very original, and there are already a plethora of other synthesizers available that can do the same, but it did give me some insight.
You see, the SID chip does not get it's magic from its oscillators; they are in fact rather simple (though I did take a liking to its bit shifting shortcut for creating noise, and I will make this method available as a waveform in Lythrum). No, the magic comes from its low bit depth in combination with its rather primitive - in regards to its physical design - filter and diode pass through. An element that the next entry in this diary shall focus on.
Determined still to find the magic in oscillators, I took a long look at the famed superwave from 90's Roland JP-8000 and started building tools that would allow me to recreate its sound. This experiment did in the end give me the sound of the superwave, but again I found that the sound did not come from the actual oscillator, but from a clever use of unison, or chorus - again something that will make it into Lythrum. What I did get out of this experiment however, in regards to the oscillator, was that I realized that the tools I had created for creating various forms of oscillators - sine, triangle, square, pulse, and all the various mixings of these, was actually far more elaborate that anything I had previously seen in the world of audio synthesis, and was closer to the tools used in "antique" analogue radio, sonar and similar very hardware based manipulation of audio waves. This, I realized, was what I had been looking for.
The depth in the sounds I adored came not from the oscillators - this came from later adjustment to the sound - but, for the uniqueness of a sound, the heart of it, the oscillator, I felt, had never really fully been explored naturally from a synth itself, and the tools I now had at my disposal, when adjusted to a more inviting interface, would enable Lythrum to do just that. I had found the heart I needed.
The interface for the oscillator tool has now been created, the code has been written, and it now only waits to be joined with the circuitry, the depth, the touch of magic that the latter effects layers will provide.
Lythrum begins opening her eyes.
As you may remember, earlier I found that much of the character of the SID chip comes from its use of simple components, and lately I've been doing experiments within the field of analogue saturation and harmonic - and to some degree inharmic - distortion, in order to simulate just this. My first course of action was to familiarize myself with circuit boards and diagrams of analog distortion and amplification units to a degree where I felt comfortable enough to simulate then in the computer. When this goal was reached, I modeled a selection of the diodes most commonly used in general amplification and guitar distortion units, since the overdrive provided by the SID chip is rather simple and somewhat easy to emulate. Though this was successful from a strictly scientific viewpoint, I found myself disappointed in the familiarity of the sound these simulations provided and in need of a different angle of approach.
I selected to pursuit various paths. First, I started looking at the harmonic distortion offered by vacuüm tube diodes and began modeling various models of these based on measurements obtained from various sources on the Internet. Secondly I began studying the harmonics, frequency spread and saturation that occurs naturally in monitors from the 60's and 70's, since this is often the sound one would associate with a classic, warm sound. Thirdly I read every interview and technical document I could find on how other developers, both individuals and large companies, had approached modeling analogue circuitry. Finally I simply looked at various methods of digitally manipulating audio, with special regards to distortion, saturation, filtering and "advanced" equalising formulas since there is a wide selection of reading material available on the subject online.
After many days of experimenting with different formulas, simulations, approximations and, I must confess, rather arcane and unusual methods of mathematical mangling of audio, I finally arrived at a method that satisfied my desire for simulating both the smooth and gentle saturation of the analogue devices of old, and the exaggerated expression that newer digital methods provide. Unfortunately, since I must try and protect the future of my products and not deprive myself of the commercial sustainability that is necessary for the continued development of this and similar products, I cannot disclose the precise method used, but I can say that it is a non-linear, transiency and frequency based method derived from a mix of the emulation of both the germanium diode and the vacuüm tube, simple wave shaping as well as an unusual digital mix of distortion, saturation and frequency modelling, combined into something that gives precisely the variation, smoothness and type of pleasant distortion I was seeking.
The result of this will be available in Lythrum, in different implementations, as both an effects unit and to achieve pleasant analogue style limitation of audio output.