An Introduction to Beamforming with MEMS Microphones

Combining multiple MEMS microphones in a beamforming array brings the inherent advantages of these compact, low-power, and cost-effective devices to applications that require directional response, such as voice-interaction systems and professional AV equipment.

Microphones and Directionality

MEMS microphones are robust, cost-effective, and easy to integrate in almost any application due to their small size and low power consumption. Their omnidirectional response, equally sensitive to sounds coming from any direction, suits certain applications such as when a fixed microphone is needed to capture sound coming from an indeterminate direction or from a moving source. On the other hand, omnidirectionality can allow ambient or unwanted sounds to compete with the main sound source making the audio less clear or difficult to hear.

A beamforming array containing multiple MEMS microphones can overcome this by boosting sounds coming from a certain direction and attenuating others. This is achieved by summing the microphone signals, using signal-processing techniques such as delay insertion, amplification, and filtering to minimize the signals from unwanted sounds. The signals representing the wanted audio source are added together, while the unwanted signals sum incoherently and are thus attenuated relative to the main signal. Figure 1 illustrates this method. The signal processing can be quite simple in the case of a carefully designed, basic two-microphone array.

Figure 1: Beamforming microphone arrays boost wanted signals relative to background noise. (Image source: Same Sky)

A basic beamforming array can contain as few as two microphones, creating an instrument that has a single axis. With suitable signal processing, a broadside array can be created to maximize the signals resulting from sounds coming directly from the side of the array, perpendicular to the axis of the two microphones. Alternatively, an endfire array is created by optimizing the directionality for sounds traveling along the microphone axis.

In each case, it is important for the microphones in the array to have closely matched sensitivity and frequency response. Fortunately, this is a key strength of MEMS microphones due to the wafer-scale fabrication processes used during manufacturing.

Broadside Arrays

Figure 2 illustrates the broadside array. Sound from the direction of the preferred source arrives at each microphone simultaneously and the outputs are summed to produce a signal of larger magnitude. Sound signals coming from other angles sum less constructively.

In practice, the broadside array is equally sensitive to sounds coming from either side of the main axis. For this reason, it is often used where little or no unwanted sound is expected to come from behind, above, or below. A typical application is to support voice interaction in a television or PC monitor, where the user is expected to be located directly in front of the screen and ambient sounds in the room are likely to come from either side rather than behind or above. The microphone array can be built into the screen’s enclosure, which allows a natural broadside orientation and is also convenient and unobtrusive.

Figure 2: The broadside array is most sensitive to sound sources located perpendicular to the microphone axis. (Image source: Same Sky)

Endfire Arrays

If audio sources from behind or beside the microphone are to be attenuated, the endfire array can minimize these while boosting the sound signals from directly in front of the array (Figure 3).

Figure 3: The endfire array can isolate sound originating from directly in front of the microphone. (Image source: Same Sky)

As the diagram illustrates, the wanted sound arrives at the first microphone and then travels a known distance to the second. Signal processing compensates for the resulting known delay and adds the two signals, producing a much larger result. Summing sound signals coming from behind the array or from off-axis produce a much smaller effect.

Typical applications for endfire arrays include handheld microphones for television or radio that are intended to be pointed towards the source – such as a presenter or speaker – to capture the speaker’s voice clearly and eliminate background noises.

Circular and Spherical Arrays

A beamforming array of, say, four or more microphones, situated on the perimeter of a circle (Figure 4) or in a spherical orientation, can provide signals that allow more complex signal-processing algorithms to determine the direction from which a received sound has originated. This type of array can be used for purposes such as intelligence gathering, including identifying the origin of gunfire in military or law-enforcement applications. Here, digital signal processing is applied to recognize the sound of gunfire as well as calculating a bearing to help pinpoint the source.

Figure 4: Larger arrays can support complex functions such as locating a sound source. (Image source: Same Sky)

Summary

MEMS microphones have an omnidirectional response and are chosen for a wide variety of applications, especially where design priorities include cost-effectiveness, reliability, small size, and low power consumption. Because they are made using semiconductor fabrication processes, parameters such as sensitivity and frequency response can be closely matched, which is an important requirement when building beamforming arrays.

Beamforming brings the strengths of MEMS microphones to applications that require directional response. An array can contain two or more microphones, and signal processing is applied to the output of each to achieve a desired directional response. Basic configurations include broadside and endfire arrays that feature relatively simple signal processing. More complex arrays include direction-finding circular or spherical arrays, or arrays comprising a handful to several hundred microphones for applications such as research or security surveillance.