Two loops, one instrument

The first path: back into the material

In NACRE, the sensor signal returns to the string through a filter and driver. The loop responds to everything: hand position, tuning changes, the gap to the exciter, and processing latency. Its purpose is to sustain the desired vibration while preserving the ability to stop it.

Draw this path as a closed arrow: string motion → pickup → filter → driver → force on the string. Every stage has an operating range, delay, and noise. If sensing is lost, the loop’s remaining energy should not lead to a state the performer cannot understand. Excitation permission and level limits therefore belong to the local device.

The second path: toward the listener

A copy of the audio signal can go to a granular instrument, spectral delay, or multichannel mixer. This branch creates an artistic continuation through the speakers. A long tail at its output does not mean the string itself is still vibrating.

Here the arrow is open: string recording → REGOLITH or MODEFIELD → mixer → loudspeaker. Mute this branch while the string continues, or leave its tail after damping the string. The separation becomes compositional material. Also check unintended acoustic return: a loudspeaker in the room can excite the body again even if the digital diagram contains no loop.

Gain alone is not enough

Phase matters in a physical loop: the returned energy can sustain motion or oppose it. Magnetic force on a steel string is also nonlinear. In a magnetically biased design, the alternating component of excitation has a more suitable relationship to the original signal; geometry and adjustment still matter.

Specify what “feedback” means in each system: audio, a pitch estimate, a control value, or an algorithm’s state may be returned. For each loop, identify what is measured, what the signal acts on, and how the action stops. This distinguishes physically sustaining a string from repeating recorded sound or reusing control events.

Why the effect stays outside

Spectral processing analyzes sound in time windows and can add substantial latency. Inside an excitation loop, that changes the conditions for sustaining vibration. NACRE therefore places external digital effects in a separate recording and monitoring branch.

Separation does not require a simple external branch. It can contain long granular layers, altered durations, and spatial movement. Its complexity should not silently alter excitation conditions. After a patch change, check the direct path, processed path, and their combination separately. This also makes it easier to explain a timbral change after a software update.

How we end the sound

The physical layer has a pedal and a threshold for reducing excitation. The digital branch has its own release and a stop for new events. In the score Sustained Contour, these endings are deliberately separated: the listener should sense which action stops each part of the sound.

A stop command needs a clear scope. Releasing a pedal may remove physical sustain without erasing grains already captured; stopping new events may leave a resonator decaying. The score identifies these states separately. Test the overall silence command on the assembled setup, including external effects, so a stage pause does not depend on an accidentally lingering tail.

A practical check

For one phrase, define four states: both paths active, physical only, processed only, and both off. Specify the transition action and any sound allowed to remain afterward. Check the same transitions with the external computer disconnected. This makes stopping and failure behavior part of the architecture while preserving the musical freedom of a long tail.