Using 3-D Microphones To Create “Quiet Models” For Aircraft

As computational power grows exponentially, researchers at NASA Langley Research Center's Structural Acoustics Branch are using the unique three dimensional sound-capturing protocol of Soundfield microphones to turn recordings into "virtual sonic landscapes" that mimic physical reality.

Dr. Ferdinand Grosveld, Dr. Stephen Rizzi and Brenda Sullivan recently completed a series of recordings using a KEMAR binaural head and a SoundField ST 250 3-D microphone system, which NASA will ultimately use to judge the accuracy of their three-dimensional models of aircraft noise. These acoustic models are expected to aid NASA engineers in designing aircraft that are significantly quieter for both passengers and observers on the ground.

NASA envisions an air transportation system based on numerous small airports as opposed to a few large ones. Internal and external noise stands as a barrier to the public's willingness to accept an air system using small local airports. To achieve theis vision, NASA needs theoretical models that accurately predict the effects of specific physical structures on sound. But first, to determine if the models they propose are accurate, they need "test cases".

To produce such a "test case", NASA's Grosveld, Rizzi and Sullivan spent several weeks investigating a vibrating plate in an anechoic chamber. The vibrating plate resembles part of an aircraft fuselage radiating noise to the passenger cabin.

They measured the physical vibrations of the plate along with its radiated sound at various locations using a KEMAR binaural head and a SoundField ST 250 microphone. Although three-dimensional sound recordings and simulations have been around for a few years, NASA's research breaks new ground.

While a virtual reality video game might successfully render a growling monster, the feat is relatively simple, because the monster is a point source. NASA's vibrating plate represents an extended surface that defies the monster's point source simplicity. Consider the next step -a whole vibrating fuselage - and you have one monster of a modeling problem on your hands!

To understand how NASA researchers will turn these recordings into three-dimensional playback, it's important to understand how we hear natural sounds in three dimensions. When a sound hits our ears, what we hear is affected by the direction it comes from. First, the distance between our ears creates a time difference, which results in a phase difference for low frequency sounds that approach us from the side.

Second, the head itself creates a shadow and time delay for high frequency sounds that approach us from the side. Our uniquely-shaped ears provide front/back and up/down information by applying unique directional filters called "head-related-transfer-functions" (HRTFs). (Place a cupped hand behind your ear in the presence of noise to hear what a profound impact it has on the frequency response of your ear.) Our brains take this information and compute the location of a sound source despite natural head movement.

The brute force approach to reproducing the binaural radiated sound requires the panel to be discretized, or treated like a bunch of smaller pieces. Each piece radiates sound like a speaker and a set of HRTFs are needed to filter the sound from each piece to the listener's right and left ears.

Depending on the complexity of the vibration, very many "speakers" may be necessary to recreate the sound. In the NASA experiments, panel vibration measurements were made at over 500 discrete points. Thus, to playback the three-dimensional sound in real-time, over 1000 filters would need to be used simultaneously and their outputs mixed.

For this reason, NASA researchers are interested in more elegant ways of tackling the problem, using SoundField's "B Format" microphone to record four channels of information corresponding to the three dimensions of space (X Y Z) plus a reference point (W). Hollywood sound designers and musicians use "B Format" to capture and then convey sound in any-channel surround systems (mono, stereo, 5.1, etc.).

NASA plans to use the four channels to recreate the binaural sound field for a listener with only eight filters - a reduction in signal processing of over 125 compared to the 1000 needed for the "brute force" method. The sound can be made to change its location by tracking the listener's head orientation and changing filters on-the-fly.

Since NASA's interests lie in developing technology for the future, it is not often possible to use an actual "B-Format" microphone because many of the proposed concepts don't physically exist. Therefore, to utilize the above scheme, Langley researchers and a team from the Mechanical Engineering Department at Virginia Tech aim to do the next best thing - predict the signals that would be measured by a virtual "B Format" microphone.

For the test case, the predictions can be compared with the actual recorded signals. Finally, subjective tests will then pit the KEMAR binaural recordings against those predicted from the filtered "B Format" microphone signals so that human beings may judge the accuracy.