Audio Surveillance Countermeasures
by Thomas A. Koenig

“Protecting Sensitive and Confidential Conversations”.

Speech surveillance of rooms where decisions are made is as old as the word “eavesdropping”. It gives an unauthorized listener not only the information he desires, but equally important, the lead time he needs to make advantageous use of that information. With modern technology, speech surveillance has expanded beyond the use of microphones as “bugging” devices.

There is an entire array of listening methods, such as laser Doppler vibration detectors, or piezoelectric vibration detectors. There are several new methods of signal transmission that are more difficult to detect, and several methods of cleaning up a noisy signal so the conversation is no longer buried in background noise. The Federal Government is cognizant of this problem and has a standard to protect itself from this kind of surveillance. They are also warning corporations that foreign governments are using their resources to pry industrial secrets from them. Most executives are totally unaware of how vulnerable their organization is to speech surveillance and consequently only a few have taken adequate steps to protect their most valuable resource: information. We describe cost-effective methods for counteracting acoustical surveillance within buildings.

The most important sound to be secured is speech, although the protection of machine sounds may be important in certain situations. The voice creates a sound pressure that propagates outward and causes very small forces on all surrounding objects, e.g., walls and windows. As a result, these objects move a very small amount. There are several listening devices that can detect such tiny movement and convert it into an electrical signal. If the signal is noisy, other devices can be used to remove some of the ambient noise to improve the intelligibility of the speech.

We can separate the countermeasures into two types: passive and active. Passive countermeasures are fixed (structural) alterations to the building that are not easy, or inexpensive to change. Walls, doors, windows, other building components, sound absorbing panels, mufflers, gaskets, caulking and any other methods which either block sound transmission or absorb the sound are of this type. Active countermeasures are controlled, variable, alterations to the acoustical environment of the building. The background sound levels and the levels of speech amplification equipment are of this type. There is a large and very significant difference between these two types: active countermeasures can be altered rapidly and inexpensively to accommodate new situations, while passive countermeasures cannot. In almost all cases, successful speech security is achieved through a well-designed combination of both types of countermeasures. Neither is generally sufficient to provide speech security by itself.

SPEECH SECURITY
The ability of an eavesdropper to understand conversations in a room depends on only three factors:

Voice Level: The sound level of the natural, amplified, or recorded voice.

Voice Reduction: The loss associated with travel from the speaker to the eavesdropper.

Background Level: The background or ambient sound level at the eavesdropper.

Speech security is obtained when the factors are combined in this way:

VOICE LEVEL – LEVEL REDUCTION = BACKGROUND LEVEL (or less)

Lowering the voice level and raising the background level are active countermeasures. Increasing the level reduction is a passive countermeasure. The passive countermeasure of room construction is most often the only acoustical variable used, and this leaves a wide gap between actual and expected room performance. For example, walls with large level reductions between a person speaking within the room and a listener outside the room are used. The sound on the other side of the wall is controlled, but we know nothing about the speech levels inside the wall. A microphone placed within the cavity of the wall receives a reduced speech level from the room, but since the ambient levels inside the wall are low, speech intelligibility can be high. Periodic sweeps are relied upon to detect such devices. If sound reinforcing equipment is used, voice levels become a design variable. If sound masking is used, the background level also becomes a design variable. The more design variables there are, the more choices for cost tradeoffs.

The sum of the Level Reduction and the Background Level is the factor that determines privacy and is under control of the designer. This is where cost tradeoffs can be made. If the Background Level can be increased, the Level Reduction can be reduced, so long as the total is still the same. If one factor is less expensive than the other, cost savings can be made with no loss in performance.

MASKING SOUND
Active control of the Background Level is called masking sound. Masking sound is more than “white noise”. It can be a combination of: broad band random sound (similar to air conditioning system noise); music; synthetically generated conversation. Random sound and music have been used for many years in the ordinary office environment, but have been used only sporadically for audio security, and have been misapplied in many cases.

Masking sound has two basic purposes: to cover or mask the speech with higher levels of sound; and reduce the ability for a sophisticated eavesdropper to use equipment that will eliminate the masking and thus recover the speech. The value of the random component of the masking sound is that it has no meaning to a listener and thus is unobtrusive, and that it covers all speech frequencies continuously in time. The value of the music component is that it can be pleasant to listen to, and that it can help to confuse the eavesdropper attempting to understand speech. The value of synthetic conversation is that it can do even a better job of confusing the eavesdropper. In the most severe applications, all three components may be used. When all three components are used together, their relative levels can be set so neither the music nor the synthetic conversation is audible to occupants of the secured room.

Masking sound as an active countermeasure has several strong advantages in addition to its cost. It can be used to accommodate the temporal changes in Voice Levels. For example, if the conversation turns heated, or an unplanned use of audio-video equipment occurs, the masking can be adjusted upward to retain security. There is no passive equivalent. When masking sound is added, the planned performance of the more expensive passive countermeasures can be reduced, resulting in a much less expensive facility. Masking sound does not require the detection, deactivation, or removal of listening devices to be effective.

VOICE LEVELS

Natural Speech
Most persons control their voice levels to permit the person to whom they are speaking to understand. In most conference rooms, this implies normal voice levels. In a larger room, and when presentations are being made, the speaker will unconsciously raise his voice to “project”. If this person is near a door, a typical weak link, this increase can have important consequences.

Amplified Speech
In board, conference or briefing rooms, speech reinforcement systems may be used. Use of such a system has several security weaknesses: the microphones have created an electrical duplicate of the conversation, which may be tappable; the resulting sound levels within the room are amplified, so they may be heard outside, recording creates a magnetic duplicate of the conversation, which may be taken; and the security of the conversations is controlled by the person running the sound system, who may not be aware of his responsibility.

Recorded Speech
Audio or video playback systems may be used. Again, it is common for these systems to be played too loudly.

EAVESDROPPING AND THE FACILITY WALLS
Walls comprise the largest part of the room surface. To provide speech security at walls, the possibility of three different types of eavesdropping must be addressed: listening on the other side of he wall by a person or microphone picking up airborne sound; listening on either of the wall surfaces, by a contact microphone or vibration probe picking up wall vibration; and within the wall cavity by a microphone picking up airborne sound.

DUCTWORK
Mechanical system ductwork (with supply and return air) can carry significant sound across the perimeter of the room. The duct material itself also can carry conversation-caused vibration across the perimeter. Usually the supply air is ducted, but eh ceiling plenum is used for return air, so a wall penetration, possibly of large size, may be created. This is a particularly bad, but common, situation. The possibility of two types of eavesdropping must be addressed: airborne sound within the duct; and speech induced vibration of he duct wall.

Duct masking devices exist; they address both types of eavesdropping, are easy to install and cause no pressure drop.

DOORS
Doors are a necessary penetration in walls. Fire codes may require not only a primary entry that is used, but also a secondary exit that is seldom, if ever, used. To provide speech security at doors, the possibility of three types of eavesdropping must be addressed: listening on the other side of the door by a person of microphone picking up airborne sound, particularly at the gaps; listening on the outer surface of the door, by a contract microphone or vibration probe picking up door vibration; and listening within the door cavity by a microphone picking up airborne sound.

Active countermeasures such as masking sound may be used to create a controlled background sound level at the door.

WINDOWS
Most executives and management personnel demand windows in their offices, despite the need for high security, and the poor acoustical performance of ordinary window. Both visual and audio surveillance is possible through windows. The possibility of two types of eavesdropping must be addressed: eavesdropping on the other side of the window by a person or microphone picking up airborne sound; and eavesdropping on the window surface itself by a contract microphone or remote vibration detector (the so-called “laser microphone”).

Active countermeasure such as a “Window Masker” can protect the window from all types a eavesdropping.

PIPING AND STRUCTURAL MEMBERS
Piping of various types can penetrate the perimeter of a secure room. Examples are halon and sprinkler piping, electrical conduit, hot and chilled water piping, and plumbing, piping. Structural columns and beams can also penetrate the perimeter. Eavesdropping is primarily through detection of the vibration caused by speech within the room and carried through the perimeter of the room. Metallic pipes or members not only carry more vibration but they can also carry electromagnetic signals beyond the room perimeter.

CABEL TRAYS AND RAISED FLOORS
Many facilities utilize cable trays in the ceiling or trenches in the floor, or raised floors for wire management. Penetrations at the room perimeter are equivalent to open air ducts and must be controlled. The possibility of two types of eavesdropping must be addressed: listening to the airborne sound; or detection of cable tray vibration induced by conversations. Masking sound again addresses both types of eavesdropping and is the most effective.

EVALUATING THE PERFORMANCE OF AUDIO COUNTERMEASURES
Once the countermeasures have been installed, they must be evaluated. Past procedures have been either subjective listening, or by making field measurements of the STC. Although necessary, neither procedure is sufficient to verify audio security.

The concept and practice of speech intelligibility is well developed. The audio security equation can be made a precise determinant of intelligibility (or lack of it) when cast into a form that takes into account the levels at various frequencies (spectrum), for either airborne sound or structure borne sound. The possibility of amplified voice levels within the room and the use of signal recovery techniques by the eavesdropper also can be taken into account. Because of the large number of calculations, it is best done with a computer.

CONCLUSION
The proper combination of active and passive measures can not only yield an effective countermeasure against audio surveillance, but it can also reduce the cost of building secure room. All of the guidelines given are based on using a combination of active and passive countermeasures to satisfy the audio security equation (lack of speech intelligibility). They are not based on meeting any specific STC criterion for a product, or on meeting any specific ambient sound level criterion.

Dynasound is a foremost Leader in the design and implementation of Active Countermeasures for Audio Surveillance. Please visit us at www.soundmasking.com.