Deep Geek: Audio Beam-Forming In The Real World

Have you ever encountered something that seems a little odd, then find that you are in fact surrounded by examples if it in your daily life. So it is with “Beam Forming.” You may never have heard of it, but it’s all around you, and it’s more than a little interesting.

Wikipedia tells use that, “Beam forming is a signal processing technique used in sensor arrays for directional signal transmission or reception.” It is essentially a way of using an array of omnidirectional sensors to synthesize directionality.

Cast into the audio domain beam forming is a way to use the signals from multiple omnidirectional microphones to create the equivalent of a direction microphone. Further, since the process is based upon signal processing, it can be variable. It can create the equivalent of an electronically steerable microphone, complete with the ability to “zoom” in or out. It’s not unlike a zoom lens for sound.

Sounds cool, right?

Why does this come to mind? It just keeps coming up in a passing way, but never with any details. So I thought it worth a little digging. Here’s more theoretical detail for the inquisitive.

For one practical example, Plantronics released a new Bluetooth headset recently called the Savor M1100. One of the things that they claim makes the M1100 special is the use of an array of three(!) microphones.

Two microphones are commonly used for noise cancellation. One microphone points at the user and the other away from the user. The background noise is present in both signals, but 180 degrees out of phase, reflecting the orientation of the microphone elements.

When the two signals are summed the anti-phase signals cancel out and the background noise is suppressed. This kind of noise cancellation is dead common, although there are various ways of implementing it.

According to Plantronics;

Two of the three microphones precisely capture the user’s voice and simultaneously cancel background noise. The third microphone automatically activates when in extremely noisy situations, helping to overcome the most challenging environments while maximizing battery life.

The remark about battery life points to one reality of such devices, the related digital signal processing techniques require processing power. The more capable DSP chips can perform more complex tasks, but also draw more power than their lesser cousins. That can have an impact on impact battery life.

Looking literally closer to me, my little HP 5102 netbook…right here on my desk…has a pair of microphones built into the lower bezel of the screen.

The audio settings in Skype offer the built-in microphone described as an “Integrated Microphone Array.” The claim is that it aids in suppressing background noise when using services like Skype and Google Talk. I’ve sometimes wondered just how effective a two element array can be?

Another example of beam forming crossed my path while I was reading about Skype’s Skype-For-TV initiative. The various manufacturers making Skype-enabled HDTVs each offer a camera & microphone module. Typically these devices have a small array of omnidirectional microphones.

The Freetalk TV Camera For Samsung TVs has an array of four omni microphones. By using some DSP techniques (delay, sum & difference) it is said to employ , “Beamforming picking voice from the camera direction.”

In this particular case the camera is not steerable, nor does it have zoom capability. That implies that the processing relating to the microphones can be relatively simple and still be effective.

More microphones increases the dataset available to the DSP algorithms, which increases the opportunity for precision in the process…albeit at the cost of adding complexity to the DSP programming.

For example, LifeSize makes a SIP conference phone called simply the LifeSize Phone. Whereas the typical conference phone uses an array of three microphones arranged around a central speaker, the LifeSize Phone employs a circular array of sixteen omni-directional microphones.

The larger set of input signals gives the LifeSize Phone great flexibility to vary it’s directivity, ensuring that background noise stays well and truly in the background.

A Scandinavian company called Squarehead Technology has taken this to an extreme, launching a new system called an “Audioscope.” Their device starts with a circular array of over 300 microphones. The signals from all of those mics are processed in such a manner as to provide wildly flexible, audio pickup that is both directionally steerable and can zoom in on distant sources.

The use case most typically cited has the device installed in a sports venue like a basketball court. The AudioScope can reportedly isolate the sound of a single individual sitting in a court-side seat, while suppressing the noise of the rest of the crowd.

One interesting aspect of the system is that it records the signals from all of the microphone channels. This allows the user to go back into the recording of the event and hunt for specific sounds, panning, tilting & zooming through the audio scene just as it it were a camera. Applications for the Squarehead Technology systems include broadcasting, large conference centers and video conferencing.

Now that I’ve laid out some examples don’t be surprised if you start finding example of beam forming all around you, too. Sometimes a microphone is just a microphone, but sometimes it’s literally something more. Sometimes it could rightfully be said to “beam you up.”

  • Acoustical reciprocity means that the same techniques can be exploited in the other direction using loudspeakers. Tall linear arrays of small loudspeakers can be phased to produce hyper directive sound fields — for instance, to distribute sound over a wide floor area while keeping it narrowly focused out of the rafters, as is done in the main hall of Grand Central Terminal in New York to keep PA announcements intelligible. The same effect is exploited in “soundbar” single-speaker home theater systems, which can simulate surround sound using a single line array of small drivers.

    • I was aware of that fact, but wanted to focus on the input end of things for this post. That has potential to impact a wide range of people.

      I must admit a certain curiosity for some “hypersonic” transducers that I read about last year. They beam a pair of ultrasonic signals at a specific location, perhaps in an art gallery just in front of a painting. Where the beams cross their interference created audible sound, but only at that very location. Sounds very interesting.