Tremulant Discussion

In the virtual pipe organ world there are currently three main methods of creating a tremulant effect.

1) A simple LFO (Low Frequency Oscillator)

The GrandOrgue and Soundfont internal tremulants are based on Frequenct Oscillation filters that are applied to the samples. This is an amplitude modulation effect, although the soundfont specification does also allow for a frequency (in cents) modulation as part of the LFO. These filters allow the tremulant effect to be synchronized across the rank, but do not model the complexities of the tremulant effect on frequency and amplitude in enough detail to create authentic sounding deep tremulants.

2) LFO Filter based on frequency analysis

The Hauptwerk built in tremulant uses LFO filters based on analysis files created from recordings of a Trem and Non Tremmed sample. This creates a closer approximation of the tremulant effect and allows for the tremulant to be synchronized across the rank. However, to some people the effect still lacks authenticity due to the synchronization of the amplitude and frequency peaks and troughs. See the tremulant frequency/amplitude analysis image below for more information.

3) Sampled Tremulants

This is the most accurate reproduction of the tremulant effect for each note, as all the effects on amplitude and frequency are correct. However, the tremulant will not be synchronized across the rank.

Amplitude and Frequency Analysis on a Viol D'Orchestra note from a Robert Morton Theatre Organ.

Complexity

The image above shows why modeling a tremulant is a complex task, and why a sampled tremulant creates the most authentic note for note reproduction of the tremulant effect. The more yellow the line, the louder the amplitude.

All of the note's harmonics have a different peak and trough amplitude location. The peaks and troughs have the same sine frequency, but their starting and ending points differ across the harmonic spectrum of the note.

Therefore a simple LFO cannot accurately model the tremulant effect.