A capacitive-piezoelectric hybrid MEMS microphone with signal fusion for enhancing signal-to-noise ratio

Better Sound from Smaller Devices

From video calls to voice assistants, tiny microphones are everywhere. Yet these components constantly struggle to pick out voices and subtle sounds from a background of hiss and hum. This article describes a new kind of chip-sized microphone that listens in two different ways at once and then cleverly combines the signals. By fusing these twin views of the same sound, the device can hear more clearly, boosting sound quality for future phones, wearables, and smart gadgets.

Two Ways of Listening on a Single Chip

Most modern chip microphones rely on either of two physical tricks. One type senses how sound bends a thin vibrating membrane and turns that motion into an electrical signal using changing electric charge between plates. The other type coats the membrane with a special material that directly generates voltage when it is squeezed or stretched. Each method has advantages and weaknesses: one can be very sensitive but noisy and hard to fabricate; the other is quieter but can miss faint details. The researchers set out to combine both sensing methods inside one tiny structure so that a single incoming sound wave is recorded in two different, complementary ways.

Designing and Building the Hybrid Microphone

The team designed a circular membrane made from layers of silicon, metal, and a thin film of aluminum nitride, a robust material that creates voltage when deformed. One part of this layered stack serves as the bending, voltage-producing element, while the silicon layers above and below act as plates of a tiny variable capacitor. When sound enters the device, the same membrane flexes, producing both a voltage in the film and a change in capacitance between the plates. The authors first built a simplified circuit-style model to predict how the mechanical motion, air flow in the tiny holes, and electrical responses interact. They then confirmed these predictions using detailed computer simulations that track motion, stress, and air pressure across the microphone.

From Computer Model to Working Prototype

Using silicon-on-insulator manufacturing, the group fabricated the hybrid microphone on a wafer similar to those used in computer chips. They carefully deposited and patterned the metal and aluminum nitride layers, etched holes and cavities beneath the membrane, and used specialized drying techniques to prevent the delicate structure from sticking or collapsing. The finished devices were mounted on circuit boards and tested in a long metal tube that provides a well-controlled sound field. By driving a loudspeaker at different levels and measuring the output, the team showed that the hybrid microphone is more sensitive across most of the audible range than versions using only one sensing method. At a common test tone of 1 kilohertz, the hybrid mode delivered the strongest response to the same sound pressure.

Cleaning Up the Signal with Smart Combination

Simply adding the two raw signals, however, did not automatically give the quietest result. The electrical path used for the capacitive part introduces extra background noise because stray capacitances force the amplifier to work in a less favorable regime. This raised the noise floor in the basic hybrid output so that it was not clearly better than the best single-mode channel. To overcome this, the researchers treated the two outputs as separate sensor channels and applied a simple form of signal fusion. They measured how noisy each channel was and how strongly their noise patterns were correlated, then assigned different weights to the two signals before adding them. Because the true sound is shared by both channels but the random noise is largely independent, the weighted sum boosts the common signal while partly canceling the uncorrelated fluctuations.

What the Results Mean for Everyday Sound

With optimized weighting, the fused signal achieved a slightly higher clarity than either sensing mode alone and significantly better performance than earlier hybrid designs. In practical terms, the microphone can detect softer sounds above its internal noise, and it does so across the typical voice and audio frequency range. This work demonstrates that building multiple sensing principles into a single tiny device, then combining their outputs intelligently, can push sound quality beyond what any single approach can offer. Such hybrid, fused-signal microphones may help future consumer and industrial products capture voices and acoustic details more faithfully, even in challenging, noisy environments.

Citation: Guan, Y., Schneider, M., Li, D. et al. A capacitive-piezoelectric hybrid MEMS microphone with signal fusion for enhancing signal-to-noise ratio. Microsyst Nanoeng 12, 136 (2026). https://doi.org/10.1038/s41378-026-01251-y

Keywords: MEMS microphone, hybrid sensor, piezoelectric, capacitive, signal fusion