Two different perspectives
An conventional microphone hears the instrument’s radiation and the room. A contact sensor detects local surface vibrations. Neither is automatically more truthful: they answer different questions.
A contact channel depends on mounting position, sensor mass, and attachment pressure. Mounting can itself slightly alter a lightweight part’s vibration. A piezo element also depends on preamplifier input impedance: an unsuitable input changes its low-frequency response. We therefore describe SEREIN recordings together with the mounting of both pickups, and HALO recordings with the conventional microphone’s position relative to the resonator openings.
Separate channels first
Record the signals separately and listen to each on its own. Mixing introduces phase relationships that depend on microphone position and sensor mounting location. It is useful to check the result in mono as well.
Aligning the first peaks does not align phase throughout the spectrum. Air and contact paths have different transfer functions; shifting a track may improve one band and worsen another. First assign roles: one channel carries the instrument’s body, another its intimate texture. Then listen to the sum, each channel, and the mono result. Save delay correction as a mix setting without overwriting the sources.
Document the conditions
Beater material, impact point, hand position, and sensor mounting all affect repeatability. A photograph of the setup and a brief note for each take can be more useful than another effect in the chain.
Start documenting the session before pressing Record. Include the individual instrument, tuning, mechanical condition, microphone positions, input gains, and sample rate. For TESSERA, add striker type and strike sequence; for SEREIN, the initial granule distribution and first tilt direction. These details help explain differences between takes.
Leave a trace of the process
Mechanical clicks and actuator noise can be part of the concept, but that should be a conscious decision. A clean reference take and a recording of background noise help distinguish an expressive detail from a technical problem.
Record a few seconds of rest before and after the action without changing gain. This helps distinguish room noise, electronics, and the object’s residual motion. Keep denoising as a separate version: it may remove quiet granules or the tail of ceramic resonance along with interference. The archive should preserve the point before deciding which sounds count as music and which as noise.
What the research adds
Hearing Touch uses contact microphones for robotic interaction. It demonstrates the value of vibration as a distinct observation channel. Our musical interpretation is that contact recording reveals interaction details while remaining different from the instrument’s airborne sound. [1] Mejia et al., 2024
Arbel and Weissman describe documenting historical woodwinds with attention to conservation and performance. We draw a connection between object, excitation method, and archival description. Museum procedures are not transferred to working instruments without considering their construction. [2] Arbel & Weissman, 2026
while fully respecting conservation constraints
A practical check
Make three versions of one short event: air, contact, and their sum. Match loudness for listening only, and note which best conveys onset, body, and decay. Repeat after one mounting change. The aim is to identify an explainable difference and assign each signal a role while retaining the option to make another mix later.
Research references
Links point to the specified arXiv versions; publication status is available on each paper’s page. The studies concern their authors’ systems. Connections to Sagittarius A Studio instruments are our engineering and musical interpretation.