Fathom's audio processor took three years to develop.
Much of this effort went into the quality of its zero aliasing oscillators and precise synchronization between its modulators and the host transport.
Fathom’s primary waveform buffers are 16,384 floating point samples in length. The processor uses Bicubic Spline interpolation between sample points at run-time to provide an extreme level of audio quality.
Spline coefficients are pre-calculated along with each sample value itself to provide the fastest possible table fetches at run-time.
Alias Free Oscillators
For an oscillator to be called alias free, the output must be constructed from partials, a series of sin waves, which control the frequency content prior to filtering. Filtering should be your choice not built into your oscillators.
Shown here is the Fathom saw tooth at 128 partials in the frequency domain. This unedited screen capture shows a high frequency noise floor well below -130 dB in relation to the amplitude, or less than one part in one million.
Notice there is no imaging or artifacts after the last partial.
When you create a waveform in Fathom all aliasing is removed no matter what you draw. We achieve this by taking your edited waveform, running it through a Fast Fourier Transform to derive the component partials and reconstructing the waveform from these partials in the time domain.
The number of partials itself can be controlled for each oscillator as well as modulated in real time.
Smooth Note Transitions
Artifacts can also occur in the time domain if a free running oscillator is started in mid-cycle or if any modulation which effects the waveform passes through the vertical edge of an envelope.
Fathom contains real time logic which enforces perfectly smooth note transitions and modulations. When Fathom's oscillator voices are detuned, the processor keeps track of the relative phase of each oscillator, and at the start of each note plays each oscillator virtually for a few microseconds until the beginning of its waveform. In this way all voices maintain their relative phase but never start in the middle of a waveform.
Shown here is a screen capture of a Fathom note transition with a single oscillator sin wave and detune set to one voice in free running mode.
If a sequencer note stops and a new one starts at the same point in the host sequence, Fathom will make the frequency transition in mid-cycle, emulating the behavior of a hardware synth.
Shown here is a note transition for a single oscillator sin wave with detune set to eight voices in re-trigger mode. All voices finish their cycle in the terminating note and start at the beginning in the new note without clipping the cycle.
This is a note transition for a single oscillator sin wave playing eight detune voices in free running polyphonic mode.
Here is the same sin wave oscillator playing eight detune voices in free running mode with a monophonic glide.