We use touch, the dominant user interface for years, to tap keyboards on laptops and tablets, to communicate with our car’s portable GPS, and to text friends and take photos from our smartphones.

Despite the great success of touch, there are other classes of applications for which an untethered user interface is a better choice. With either tiny screens — or no screens at all — wearables are an ideal match for touchless applications. Designers need to know about the enabling qualities of micro-electromechanical systems (MEMS) microphone arrays and ultrasonic sensing chips — two products that are ushering in hands-free user interfaces for wearables.

March of the Speech Assistant

Figure 1. Featuring MEMS microphones and a MEMS accelerometer, Bragi Dash is among the first set of “hearables” to use voice as the primary user interface. (Credit: Bragi)

Pervasive and excellent speech assistants from Amazon (Alexa), Google (Google Now), and Apple (Siri) have laid a software foundation that good hardware can leverage. Guess Connect is a new Alexa-enabled smartwatch, and other Alexa-based wearables are in development. Google Now is in Android Wear smartwatches and fitness trackers from companies such as LG, Polar, Huawei, and Tag Heuer. Apple uses Siri in its Apple Watch and AirPods, its smart wireless earbuds.

Smaller competitors are offering premium functionality in their earphones. Bolstered by its third-party speech assistant, NeoSpeech, Bragi Dash (see Figure 1) includes buttons for simple functions such as answering phone calls or changing the volume in its smart earphones.

Doppler Labs calls its HearOne earbuds “the world’s first in-ear computing platform.” HearOne supports both Siri and Google Now, allowing Apple and Google fans to continue to use familiar speech assistants.

The common critical component in these wearables and hearables is the MEMS microphone. Because the devices reside at the start of the audio signal chain, improvements in MEMS microphones can have dramatic effects on the voice-assisted user experience — thereby making the audio components a point of differentiation for designers.

Two Microphone Architectures

Figure 2. InvenSense TDM microphones remove the need for ADCs, reducing package size and cost while supporting microphone arrays — an important feature for wearables such as hearables and AR/VR eyewear. (Credit: InvenSense)

There are two types of MEMS microphones: capacitive and piezoelectric. Knowles Electronics, the largest supplier of MEMS microphones, released the first capacitive devices in 2000.

Capacitive MEMS microphones operate on the same physical principle as the condenser microphones that have been around for generations. The technologies measure sound with a capacitor, or condenser, made of a flexible diaphragm and a rigid backplate. The diaphragm vibrates in response to sound waves, with bias high enough to create a signal. Capacitive MEMS microphones are by far today’s dominant architecture.

Piezoelectric MEMS microphones, which just reached commercial markets in late 2016, operate upon a different physical principle. Leveraging an energy-harvesting material to generate a signal, piezoelectric MEMS microphones directly convert mechanical sound energy into electrical energy.

In piezoelectric MEMS microphones, the backplate and diaphragm are replaced by a single layer of flexible plates. Changes in sound pressure cause the plates to bend and experience stress. As the plates are built from piezoelectric materials, the stress generates electrical charge, which enables the direct measurement of sound.

Evolving MEMS Microphones

Figure 3. Vesper’s piezoelectric MEMS microphones operate in tough environmental conditions, including being immersed in dust for eight hours (left) and boiled in corn oil (right) (Credit: Vesper)

Dedicated to advancing the voice user interface through their respective technologies, capacitive and piezoelectric MEMS microphone suppliers are closely attuned to the migration toward a voice user interface. Many manufacturers see potentially game-changing enhancements in MEMS microphones that will affect wearables and hearables.

Nicolas Sauvage, senior director of ecosystem at InvenSense, a capacitive MEMS microphone supplier headquartered in San Jose, CA, considers power consumption a top priority, as new technologies require always-on use cases (see Figure 2).

Wake-on-sound capability in some MEMS microphones allows designers to add always-on voice interfaces to any connected device — without increasing power consumption. Through wake-on-sound MEMS microphones, battery-powered devices, such as smart speakers or TV remotes, stay in sleep mode until they are awakened by a noise. The device can then respond to voice commands, freeing users from having to push a button to interact with the device.

“A device is always listening to the user’s voice trigger and commands, rather than having to be close to the device and manually press a button,” said Sauvage.

According to Matt Crowley, CEO of the Boston, MA-based piezoelectric MEMS microphone company Vesper, such a “zero-power” listening functionality will allow batteries to last much longer, extending the battery-recharge time from hours to days or weeks (see Figure 3).

“That’s going to remove a major adoption barrier for hearables and wearables,” said Crowley.

Despite many advancements in software, including voice-to-text transcription, natural language processing, and user identification, however, there is still room for improvement in hardware.

“The quality of the microphone’s signal is still limited by mechanical microphones that are prone to mechanical failure modes — and this is a big issue for arrays,” said Crowley. “The failure of a single microphone can cause an entire array to malfunction.”

Immunity to common failures such as stiction, particle contamination, solder flux contamination, liquid ingress, and breakage lays the foundation for more reliable and robust microphone arrays. Wearables and hearables designers must therefore gravitate towards more rugged microphones.