Sound masking is the addition of sound created by special digital generators and distributed by normally unseen speakers through an area to reduce distractions or provide confidentiality where needed. The sound is broad band random that conveys no information about itself to a listener. It is often referred to erroneously as white noise or pink noise; the sound spectrum and level is specially shaped to provide the degree of privacy desired by occupants. Masking operates by covering up or masking unwanted sounds, similar to one-way windows that block the ability for a person to see persons behind them, or perfume that covers up other body odors. This is in contrast to the technique of active noise control which attempts to eliminate the unwanted sound. Sound masking is used in homes, commercial offices, medical facilities, court rooms, and in secure facilities to provide secrecy.
The Need for Sound Masking
Based on Sound Masking Done Right.
Effects of Noise on People
A seminal work covers the subject in some detail. Noise is defined as unwanted sound. It can have three effects depending mostly on level. At high levels, there are mechanical changes in a person, such as heating of the skin, rupture of the eardrum, or vibration of the eyeballs or internal organs. At lower levels, there are physiological (biological) changes in a person, such as elevation of blood pressure, or stress. At still lower levels, the changes are psychological (subjective) such as annoyance and complaints. Annoyance is based on factors such as the person's evaluation of the necessity of the noise, or whether it can be controlled, or whether it is normal for the environment (see Common Opinions about Sound). The levels of sound masking are sufficiently low that they have no known physical or physiological effects on people. One aim of sound masking design is to make the sound be "normal", i.e., acceptable.
Studies of Noise in the Office Environment
Since most sound masking is used in offices, a number of cognitive psychology studies have been made that relate specifically to the office environment. One study found that there was a modest stress (physiological) increase and diminished motivation caused by typical office noises, including speech. It is recommended that the use of sound masking is under the control of the worker. Another study suggested that changes in level are an important factor, but that habituation to the noise can occur. In the office, habituation can be interpreted to mean "I’ve grown used to the noise and it no longer distracts me" or "Since I cannot do anything about it, I will have to live with it." Another study point out that the specific information within the speech intrusion is not important nor is the "intensity" (level) of the sound between 48 and 76 dBA. Since the energy level of the louder sound was 1,000 times that of the least, one must assume that distraction occurred for all levels. For arithmetic tasks, both speech and non-verbal intrusive noises caused significant performance decreases. For "prose tasks" it was found that speech caused a greater performance decrease than nonverbal noises. In another study the author added several significant observations. It was found that "during a serial recall task, the accuracy of report decreases 30 to 50%." When the intrusive speech was increasingly filtered to a meaningless mumble, there was a monotonic increase in performance. Finally, the author states: "Perhaps the single feature that makes the irrelevant speech phenomena so fascinating is that the processing of sound is obligatory; it appears beyond the individual’s control." Within the references cited above are further references to earlier works on this subject. There are several implications for a sound masking system. The masking must reduce the difference between the steady background level and the transient levels associated with both speech and other sounds. Motivation and productivity are improved when this is accomplished. The masking sound itself must not change rapidly and should be as meaningless as possible.
Common Opinions about Sound
Sound masking must satisfy the persons that listen to it. People ask themselves a number of questions about the acoustical environment. The following questions were deduced from employee comments about their office environment. These are questions the listeners implicitly ask themselves to determine their response to their environment. The design of a sound masking system must take these opinions into account.
- Is the sound made by me or made on my behalf?
- Is the sound "normal" for this environment?
- Is the sound necessary and can anything be done to control it?
- Does the sound have meaning?
- Is the sound frightening?
- Will the sound have an adverse effect on my health?
- What is the pitch of the sound?
- How reverberant is the room?
Complaints about Noise
Since sound masking is a shaped random sound, often erroneously called "white noise", It is important for a sound masking system to dispel that this sound is actually noise. The most important finding by Kryter is encapsulated by this statement:
- "The general finding that the performance of the more anxious personality types is more affected by noise than that of nonanxious types would attest to the existence of a stimulus-contingency factor. In terms of learning or conditioning, the task becomes disliked and is performed relatively poorly because it is related to or contingent upon the aversive noise."
- "A possible teaching of much of the data presented in this book is that, other than as a damaging agent to the ear and as a masker of auditory information, noise will not harm the organism or interfere with mental or motor performance."
In well designed sound masking systems, distraction caused by extraneous conversations (loss of privacy) far outweighs any negative response to the sound masking
The Quest for Quiet
The biggest problem with sound masking is that when people are annoyed by the activity sounds around them (noise), they search for "quiet." They believe that "quiet" is a desirable condition of low background sound level, but what they are really searching for is the freedom from the acoustical distractions that ultimately cause annoyance. The only way to achieve true quiet would be to maintain a low background sound level with no transient sounds; a condition that requires complete isolation from all activity sounds. A better definition of "quiet" would be the absence of distracting sounds, not the absence of all sound. This is the definition used in sound masking.
Steady vs. Transient Sounds
Steady sounds are the background sounds in any environment that are reasonably continuous and long term. If a steady sound persists for a long time without change, and the level is relatively low, persons generally accept it as normal. People are seldom aware that an outdoor background sound exists. Steady sound can be tonal or random. If the sound is tonal it will create more annoyance than a random sound of the same level since the latter conveys no information to the listener. Transient sounds are conversation, paging, machine sounds as well as exterior sounds such as passing aircraft and road traffic. They are short term, can vary considerably in level, and generally distract a person's attention if the level is high relative to the steady sound level (a rise of about 10 dB is a common criterion). The distraction is further strengthened if the sound has high information content, such as conversation. At relatively low levels, the major concern is the psychological effect of distraction and annoyance. The primary use of sound masking is to reduce the distraction associated with transient sounds, and in some cases reduce the intelligibility of those transient sounds (closed offices, secure facilities).
What is sound masking?
The word "mask" means merely to cover up, or disguise, something. That something is not changed, but simply hidden. Physical masks cover the face of the wearer. Deodorants mask odors; they do not eliminate them. One-way windows mask the persons on the other side so they are not visible. Sound can mask other sounds to cover them up. In every case, the objective is to hide something that exists, it does not eliminate them. Many people have heard of noise cancellation, but incorrectly believe that it is a form of sound masking. In noise cancellation the sound is actually eliminated not covered up. So why not use it? Unfortunately, this technique works only within spatially constrained areas, such as headphones, and is not applicable to entire rooms. What is sound masking trying to mask? It covers up distracting sounds, such as conversations, by raising the background level. That level must be above the distracting sound most of the time. How is that determined? For every situation there is an optimum background sound level, just as there is an optimum light level. In the home that sound level is low, and at sporting events that level is high. The function of sound masking is to bring the background level up to the optimum; the level that provides fewer distractions without being a distraction itself. This requires persons experienced with this technique to find that optimum.
The Early History of Sound Masking
It is likely that primitive people did not want acoustical privacy, so they never camped near a rushing stream. They understood that stream noise would mask the approach of enemies or predators. The sound of fountains in Roman villas certainly served to mask the sounds of iron-rimmed chariot wheels on the cobble-stoned streets. Fountain masking has carried over to shopping malls or buildings with large atria. There are stories of a dentist, in the 1940s, applying random sound to patient’s ears through earphones to mask the terrible noise of slow speed drills. An example of a self-contained masker, made in the 1960s, is shown in the figure on the right. The application of electronics and the advent of the open office resulted in the rapid evolution of sound masking. The Quickborner Team of Germany introduced the concept of the open office to the United States in the late 1960s. Geiger-Hamme Laboratories developed a standard for open office acoustics in the 1970s. It was sponsored by the Public Building Service of the General Services Administration for use in open government offices. It included a requirement for sound masking. Many of the major furniture manufacturers, such as Herman Miller, Steelcase, and Haworth, converted much of their production to products for the open office. Herman Miller was the first to have self-contained maskers mounted pointing up on top of open office furniture panels. It did not survive primarily due to the presence of controls available to employees. Owens Corning, a manufacturer of fiberglass products, entered the open office market and introduced a centralized masking system using a speaker called the Sweeny baffle. This speaker was unusual in that the sound spectrum on axis was the same as the electrical spectrum. Unfortunately, off axis the spectrum was different; it no longer exists. Manufacturers of commercial sound systems entered the market in the 1970s; masking was an add-on to their other audio products. Soundolier (now part of Atlas Sound) sold a self contained masker that has survived until recently. The Dukane Corporation sold a masker that had two speakers contained in a heavy triangular enclosure. Companies that considered sound masking as their primary business came into existence about that time. One product, the Lahti masker, was a speaker mounted on the surface of a plastic sphere. Dynasound, Inc. introduced a masker that had a speaker mounted on the lid of two gallon paint can; the handle was used for plenum mounting. K.R. Moeller Associates sold a self-contained masker called Scamp, while the Lencore Corporation sold an equivalent unit, now called Spectra. A document was published in 1980 by the Defense Intelligence Agency. It concerned protection of secure facilities from deliberate audio surveillance; sound masking was one means of protection. Dynasound, Inc. developed a vibration device that could be attached to various surfaces such as doors, walls, and windows, to provide sound masking. In the 1980s, the Bertagni family developed a speaker that could not be distinguished from a fiberglass ceiling tile. The invisibility aspect was favorable to architects. The attempt was made to use this speaker for sound masking, but the cost and the sound radiating characteristics limited its use for that purpose. Armstrong World Industries, a manufacturer of ceiling materials, developed a similar speaker. This speaker has survived as a product of Sound Advance and is now used for applications other than sound masking. An early attitude among owners, designers, and architects was that masking was an excuse for a bad open office design. In early systems, the installers were not knowledgeable about how to provide privacy. As a result, many systems merely provided more noise and were shut off. This was countered by the rise of firms who specialized in the design, installation, and equalization, of masking systems. The evolution of sound masking since the 1980s is described in other sections.
Why Sound Masking is used
- Sound masking is dynamic (variable) as are the sounds it is intended to block. Building elements that provide sound attenuation are static (fixed) and cannot adapt to intruding sounds that change in level. Masking can vary from location to location as well as from time to time, to adapt to changing environmental conditions.
- Sound masking is by far the least expensive tool for providing privacy.
- Sound masking systems are special audio systems that create spatially uniform sound levels. The uniformity can be used to integrate music and paging into a system.
- Sound masking works at the listener's ear and is independent of the building structure (acoustically) so concerns about how distracting sounds get from one place to another is unimportant. Structural modifications require such knowledge and are more costly.
- It helps to overcome neighborhood noises, snoring sounds from other family members, It provides soothing sleep inducing sounds for babies, afternoon and day time naps are more comfortable; it creates a noise free environment for reading or studying. It provides a personalized and discreet environment for confidential conversations. There are two groups of people that should not be exposed to sound masking: those with significant hearing loss and those with very limited vision. The first group already have considerable privacy which results in masking providing too much privacy. Visually impaired persons use acoustical cues to navigate; sound masking can remove those cues.
Applications of Sound Masking
Closed offices and conference rooms often appear to provide confidentiality but actually may not. Lightweight, or movable, walls are more sound transparent and most do not extend to the ceiling deck, so speech can pass into the ceiling plenum and then to the next office. Sound masking can be used to make up for such acoustical weaknesses.
Open offices can have a background sound level that is too low. Sound level restrictions of air handling units are increasingly stricter. The conversations of others can be clearly understood. Sound masking raises the background level to make up for the absence of walls that would otherwise block the sound.
Less Common Applications
Although the above applications relate to acoustical privacy within an office, there have been applications where sound masking was used to create privacy from sounds exterior to the office. Examples are privacy from elevated freeway traffic, continual siren use in cities, sound from the floor above, and local construction noise. Sound masking is used in court rooms to prevent jurors from hearing attorney conversations with the judge at his bench.
The Congress of the United States passed the Health Insurance Portability and Accountability Act (HIPAA) into law. It mandates that individually identifiable patient health information be protected. Although written and computer files are obviously to be protected, verbal information must also be protected. "Covered entities" (those who must comply with the law) must make reasonable efforts to safeguard patient information from being overheard. The law itself gives no specific guidance on how this is to be accomplished, but a document released by the Department of Health and Human Services provides some clarification. It includes, as part of the protection, the phrase "health information whether it is on paper, in computers, or communicated orally". The Office of Civil Rights also has published a document on this issue, stating that the law does not require retrofitting spaces, such as soundproofing of rooms, in order to comply. As a result, many medical facilities have already realized that compliance is wise and have begun retrofitting their facilities. Experience has suggested that most hospital rooms do not need sound proofing, but can benefit from sound masking.
Noise in hospitals has been a problem for at least fifty years, in part because of the need to have all surfaces hard and cleanable. A large number of measurements and reports in prestigious journals have established the problem From the patient’s viewpoint the problem has been the distraction and annoyance caused by the noise of people, which results in less rest, poorer sleep, and possibly longer recuperation time. The increased socialization now permitted in hospitals, as well as the increased use of medical machinery, has exacerbated the problem. An extensive survey by the Public Health Service in 1963 showed that patients were frequently disturbed by speech and distress sounds in other rooms as well as staff visits during night hours. Other studies concerned the interference of sleep and recuperation by noise. One study found that the amount and rate of increase in the sound level from the constant background was the main contributor to full awakening or changes in the stage of sleep. It was determined that the magnitude of the change in level, regardless of its median value, was more significant than the level of a steady sound of the same median value. This conclusion was supported by an Environmental Protection Agency document. Suter expanded this finding by stating "it is clear that intermittent and impulsive noise is more disturbing than continuous noise of equivalent energy, and that meaningful sounds are more likely to produce sleep disruption than sounds with neutral content." These conclusions were the same as those found for open offices.
Sound Masking in Medical Facilities
The finding of researchers has shown that the privacy problem in medical facilities is very similar to that in open offices. As a result, sound masking has been used beneficially in a number of locations. Patient rooms, corridors, and nursing areas of hospitals are prime locations. Sound masking can be used in retirement and rehabilitation centers. It is also beneficial in medical suites or patient contact areas of medical insurance providers. Pharmacies can also provide confidential privacy at contact areas with sound masking.
Secure facilities require more care when sound masking is used. For commercial offices and even for medical facilities, the listener for confidential conversations are presumed to be casual or accidental. For secure facilities, the listener is presumed to be deliberate and may make use of sophisticated technical listening devices. Many government facilities have made use of structural solutions, i.e., rooms (room-within-a-room) that are shielded from vibration, acoustical, and electromagnetic surveillance. Unfortunately, not all secret conversations take place in such rooms. A less obvious weakness in secure rooms is that modern listening devices can be placed in locations that the building structure cannot protect against (inside wall cavities or remote detection of window vibration). Another weakness in rooms of this type is that designers may presume speech is on a controlled, but low, level. Public address systems, speaker phones, and audio/video presentations require additional protection.
There are many unclassified government documents that define how secure rooms are to be designed. One type of facility is called a SCIF (Secure Compartmented Information Facility); there are others. There are likely classified documents as well. In almost all of the documents, sound masking is one recommended audio protection method. There is similar standard that applies to financial institutions.
Categories of Surveillance
There are two categories; each must be addressed differently. Uncontrolled areas are those where the persons attempting to protect themselves have little or no control over the surrounding environment. This might be all areas outside the building in which the secure room resides, such as parking lots or other public spaces where it is possible to gain access without detection. Controlled areas are those within the building where the occupant has a measure of control.
Types of Masking Signals
Unlike sound masking in offices or hospitals, it is necessary to consider that sophisticated listeners may have technology to recover speech buried in masking sound. To inhibit such devices, it may be necessary to provide layered audio protection; several different signals are mixed together. Non-stationary random noise should be the first layer; it provides more protection than standard sound masking generators. Music may be used as the second layer; it is buried below the random noise so it is actually inaudible to room occupants. A voice babble generator or actual speech samples may be used as a third layer. The fourth layer, the actual voices to protect, should be sufficiently buried.
Types of Masking Speakers
Standard masking speakers, discussed in Section 2.1.3 must be supplemented with vibration maskers for attachment to a number of surfaces, as noted below.
Locations for Protection
Windows generally face uncontrolled areas (areas not under the control of the secure facility). They require protection. Most persons are not aware that windows respond well in the speech range of frequencies. The vibration caused by room speech is minute but can be detected remotely by laser microphones or directional microphones. The laser microphone system transmitter sends an infrared signal to the window and a special detector picks up the reflection. The reflection has been modified by the window vibration, so when the base frequency is removed (much the same as in radios) the detected vibration is that of a conversation or other internal sound. The location of the detector must be carefully chosen but the distance can be quite far. The directional microphone detects the very low level sound emitted by the window. The emitted sound spreads out, so the microphone angular position is not critical, but the distance to the window must be reasonably close. For protection, a vibration device is attached to the window that causes the window to vibrate with an appropriate masking signal. It is used to protect against vibration detection and radiated sound detection. The figure on the right shows the relative sound levels. The sound emitted within the secure room is insufficient to cause any interference with normal level conversation.
Exterior walls are generally constructed of heavier material than interior walls and seldom need audio protection. Interior walls are a different story. Speech can excite a wall to vibrate and there are several locations from which conversations can be detected:
- remote from the wall with laser or directional microphones.
- on the far side of the wall with vibration detectors or direct listening.
- within the wall cavity with vibration detectors or fiber optic microphones.
Vibration sound masking devices provide protection against these forms of surveillance when placed within the room to be protected and at the appropriate height and spacing on the wall, In effect, the wall becomes a masking speaker. A diagram of the sound levels in the figure on the right suggest that the masking protects the wall panels, the wall cavity, and beyond the wall, while the masking levels in the secure room do not interfere with speech.
Doors are weak links in walls; they may be hollow or solid core, metal, or specially built for high sound attenuation. They can open to exterior uncontrolled areas or to internal controlled areas. Every door has a gap around its periphery that may have gaskets. Because carpeting is often used, there may be a significant gap at the bottom. Most surveillance of internal doors is accomplished by direct listening while other methods may be used for external doors. Because of the multiple paths of sound through and around a door, vibration sound maskers are used on doors. The figure on the right shows one such installation.
Listening through air ducts is a time honored method of eavesdropping since almost all modern rooms have supply ducts that connect to a multiplicity of rooms. Ducts can be effective speaking tubes; the speech attenuation is particularly weak in unlined metal ducts. In some cases, exhaust ducts will connect to uncontrolled spaces. Surveillance can be accomplished by direct listening or with probe microphones or vibration devices within the duct where they are not visible for inspection. Vibration sound maskers, appropriately located at room perimeters, are attached directly to metallic duct walls and use the duct wall as the masking speaker. The figure on the right shows the masking of conversations in the duct afforded by a vibration masker at the room wall. External speaker maskers must be used at appropriate positions to radiate sound into fiberglass ducts to provide the same protection.
Normally, liquid filled pipes do not carry significant speech energy nor does conduit piping filled with wires. However, empty conduit pipes are excellent speaking tubes; vibration maskers are attached to them. In some facilities, vibration maskers have been attached to support pipes and columns.
Raised floors are generally inaccessible for inspection so surveillance can be accomplished there. If the floor is part of the air handling system, probe microphones or fiber-optic microphones can be effective. If the floor is continuous, the high sound attenuation of the material makes surveillance with vibration detectors more advantageous since the stiff metal plates respond to speech. Although vibration maskers can be attached to the floor plates, the preferred protection is to use speaker maskers under the floor.
Secure rooms often have a suspended ceiling with a plenum above. Sound masking penetrates the ceiling material in open offices, so speech can go in the opposite direction into the plenum. It is likely that the perimeter walls extend to the structural ceiling. If the plenum is part of the air handling system or if there are cable run penetrations, probe microphones can be used for surveillance. If not, a small penetration can be made to insert these devices. As with commercial offices, plenum speaker maskers are installed.
Many building codes require the presence of speakers in a secure room for emergency announcements. Although speakers are intended for creating sound, the speaker cone also responds to external sound and the coil generates a minute voltage characteristic of that sound. With proper sensing, that voltage can be converted to speech. Although it is possible to place a speaker masker next to the paging speaker the recommended solution is an optical isolator. It is essentially an audio diode; sound only goes one way.
There is some evidence that the sound of key strokes can be detected to identify the characters being entered. A vibration masker placed under the keyboard will radiate sufficient masking to block surveillance without disturbing the user.
Sound masking generators have been used for personal application for many years. There are several manufacturers of devices for home, hotel, or air travel, primarily for sleeping. The figure on the right shows a masker that has been on the market for many years.
Status of Sound Masking
Advances in Sound Masking
The advent of sound masking made use of various components typical of other types of sound systems, such as amplifiers and loudspeakers. Since that time, use of masking has grown so that manufacturers have added a number of functions to their systems that are beneficial to sound masking. Several are listed below. Chanaud has given a more detailed discussion.
Initial Ramp Up Function
On the initiation of a sound masking system, it is important not raise the background level experienced by occupants from a low existing level to a higher masking level. This function permits the level to be raised slowly and automatically over periods as long as 30 days.
Fast Ramp Up Function
Buildings will have power failures shutting the masking system down. This function prevents the level from jumping up when power is restored. Typical recovery times are in minutes.
Programmed Level Control
The need for privacy varies throughout the day. Persons desire privacy during busy times, but do not need as much when occupancy is low in the evening or on weekends. Security guards do not want privacy as they patrol an office at night. This function automatically and continually alters the overall sound masking during the day. The time history can be different for each day of the week and for each channel of masking so can be used for both open and closed offices. It requires pre-knowledge of the activity in the office to be set correctly. The figure on the right shows an example commonly used for open offices..
Programmed level control must presume daily activity levels in an office. In many cases this is adequate. However, in the 2000s, Soft dB was the pioneer of adaptive masking control. This function senses the actual activity levels in real time and automatically adjusts the masking level to continually minimize distractions. The figure on the right shows a time history of activity levels (ambient noise) in an open office and the masking sound response to that activity. Changes in masking level must be sufficiently slow that they are not noticed by occupants.
Older masking sound systems required visits for major adjustments after initial setup. Newer masking systems now have the capability to be adjusted remotely, either locally or over the internet. Because of this added capability, these systems have software that can adjust the masking level in zones, small groups of speakers or even individual speakers. Some also can adjust the masking spectrum to this level. Some of these systems have an available system zone map that identifies those zones or speakers that are to be modified. This function eliminates the need for older monitor panels, simplifies the facility managers tasks, and the expense of contractor visits.
The performance of sound masking as a privacy tool is determined by the proper setting of the system level and spectrum (equalization, tuning). Improper tuning has been the cause of system rejection in the past. One reason was that the level and spectrum in a zone was set in an iterative manner. One person measured the spectrum and another adjusted the various frequency bands of the generator manually, a slow and unreliable process. Software has been developed that allows a number of spectra to be measured in a zone and appropriately averaged. The average can be uploaded to the generator and internally compared with the desired spectrum. Then the generator automatically adjusts its electrical output to provide the correct acoustical spectrum in the zone. Uploading can be done by physical connection or remotely.
Attributes of Successful Sound Masking Systems
There are a number of factors that contribute to the success of a sound masking system. Good design takes into account each of these factors; a system with more of them will survive longer.
Sound Masking Level
The system should be able to generate sound that masks the intelligibility of speech for the various degrees of privacy. It should also be able to mask the sound of aircraft, the sound of vehicles, the barks of dogs, the music of neighbors, and other sources of annoyance should the need arise. To do this, the equipment must have a broad range of levels and an adequate number of zones.
Sound Masking Spectrum
The system should be able to apply different masking spectra at different locations. To do this, the system must have an adequate number of channels to set the needed spectra. Speakers in open offices, closed offices, and vibration maskers all require different spectra.
Spatial uniformity of sound masking in, or between, closed offices is not generally a concern. How is spatial uniformity in an open office defined? There are two ways: (1) require that the overall A-weighted sound level be reasonably constant over a specific area, and; (2) require that variations in the masking spectrum be minimized. Most successful systems use the former requirement; strict spatial uniformity of the spectrum is exceedingly difficult to achieve except when phasing is of concern (see Phasing below). Uniformity of sound masking is only important in areas that are geometrically and uniformly similar. In some zones, panel heights are different and the requirement for uniformity of speech privacy may suggest masking level differences. Also there are conditions that suggest deliberate level changes (See Soundscaping below). Deviations from uniformity must be done carefully. Early requirements were for detailed measurements in a number of workstations that showed the level variations were below a certain number, such as +/- 2 dB(A). Most workstation occupants are more concerned with the sound from others and are willing to accept some level variations within their own area. Successful systems are those in which the aisles where persons walk meet uniformity requirements; changes there occur more rapidly and so are more noticeable. Uniformity is best accomplished by proper placement of the masking speakers during installation and inspection and adjustment of the equipment prior to occupancy.
In early systems, digital masking generators had noticeable short term levels changes; the cycle time was so short that the same sound was created repetitively. As a result, some specifications required temporal uniformity in the short term. This problem was bypassed with the requirement for use of analog generators that created truly random sound. The problem has been designed away so almost all masking generators now create a digital signal that can only be interpreted by a listener as random. Now, masking sound in the short term has a constant level. The same is not true for longer periods. The privacy needs of occupants vary throughout the workday and the only variable that can accommodate this is sound masking. During hours with high activity levels, the person would need more masking to maintain his or her privacy. At low activity levels, such as early evening hours, the person would like to be aware of the presence of others; the need for community partially outweighs the need for privacy, so masking levels need to be lower. If the building has roving security personnel after working hours, there is no need for privacy and the sound masking level should be minimized. Successful systems have a clock and a control function that will continually vary the masking level in small increments over twenty four hours and for each day of the week. There are two types. One in which the level variations are programmed in advance and one in which the actual activity sound levels are measured and the masking level is set automatically to create the needed privacy. See the programmed level control and the adaptive function listed above. Unsuccessful systems had an ON-OFF clock-controlled switch which is not recommended.
When two adjacent speakers are wired to create identical masking sound, a listener passing by the midpoint between them may detect a "swishing" sound. It relates to phase relationships at various frequencies. This negative effect is most noticeable when speakers wired this way are placed in a plenum above a fiberglass ceiling tile, or when placed face down in an suspended ceiling. Successful systems avoid this negative effect. In most cases with lower NRC or higher CAC ceilings, or beneath a raised floor, tree wiring that branches out the same signal to all the speakers is acceptable. If there is concern about this effect, the wiring is changed so two independent channels of masking are fed to a checkerboard wiring array such as shown in the figure on the right. Alternate speakers have a masking signal that is unrelated to the others. Inspection is necessary to insure the wiring is correct. In some recent systems, the wires connecting each speaker are cables that carry several channels of masking and the channel feeding each speaker is chosen to avoid this effect.
Lighting engineers for years have had to contend with glare, the visual discomfort caused by direct lighting. They developed Equivalent Sphere Illumination to measure it. They then developed Visual Comfort Probability. High values of comfort imply more diffuse lighting; the light comes from many directions. That concept has never been seriously applied to sound. Most people continually experience ambient sound and are seldom aware of it if the level is reasonable; it is often referred to as background sound. That is because the sound they hear comes from many sources, the sound field is diffuse. This concept applies to sound masking as well. Experience has shown that when an office occupant can point at the source of the sound masking (the sound field is more direct), acceptance is diminished. Diffusion of masking sound is best accomplished by the presence of intervening sound attenuating materials, such as ceilings and floors. Of the locations listed above, under floor masking creates the best diffusion.
Broad Range of Equipment
The system manufacturer should have a sufficient number of types, sizes, and shapes of speakers available so it can accommodate a wide variety of masking applications, possibly on the same project. The system should be able to handle indoor and outdoor applications, large, small, or no plenum ceilings, and large or small access floor cavities. Both loudspeakers and vibration devices should be available. For visible masker locations, speaker shape and color must be acceptable to the owner. The system should be able to incorporate paging and music; it increases system utility and provides an economic advantage to the user. The system should be able to equalize the spectrum and level at least in each zone.
Sound masking must be acceptable to the listener. Masking that is truly background is most acceptable see below). Incorporating advanced functions, noted above, improves user acceptability. Masking should accomplish the privacy goal by having acceptable levels. This is a more difficult factor as the system installer seldom has control over the other two factors that determine privacy. Information exchange between team members during design helps to improve acceptability. Realistic privacy goals are critical. When the other privacy factors cannot be combined with reasonable levels of sound masking to achieve the desired privacy, the privacy goal must be changed.
The masking sound should be truly background so it does not call attention to itself. the masking spectrum should be random and neutral, not too much low or high frequency sound. Level changes, if desired, should occur slowly. Spatial uniformity reduces noticeability. The equipment should not be noticeable; this includes general inaccessibility to controls and invisibility of speakers.
The system should be centrally controllable either manually or remotely. All controls must be available to the installer; a limited number of controls should be available to the owner, and none to employees. The controls available to the owner must only be those to change level. They must be stepped (not analog) and have a small step and a limited level range so the correct functioning of the system is retained. Manual controls are acceptable if the cabinet or room is lockable. A computer software interface with the controls is preferable since passwords can be used.
Masking system equipment costs should be comparable to those of other sound systems. Equipment should be designed to minimize the cost of installation.
The system should be readily expandable. Centralized systems should have excess cabinet space, spare zone controls, and adequate power capacity. Distributed systems require only correctly located power outlets for expansion.
The system must comply with local building codes as well as national and international standards.
In most environments, the level of sound experienced changes gradually as one moves from place to place. Gradual change is more acceptable than rapid change. The same is true of sound masking. Persons moving from masked areas to unmasked areas will notice the change in level as they walk. This typically occurs when moving from an open office area to a corridor or to an unmasked support area such as for printers or copiers. This event violates the design rule that sound masking should not call attention to itself. This event can be avoided by changing the sound level more gradually. It is done by adding a string of speakers, each of which is successively reduced in level to the unmasked background level.
Masking System Types
There are several types of masking systems.
- Portable systems. These systems are small and can be installed on a temporary basis. They are controlled and powered centrally. They are used for Secret Privacy in rental rooms, and can be used for demonstration purposes elsewhere.
- Self-contained systems. These are typically one, or several, single units that contain all the features necessary to create masking sound. Each unit is powered by standard line voltage and there is no centralized control, but they are portable. They are used in homes, on desk tops, and can be carried in luggage for privacy in hotel rooms.
- Centralized systems. They can be of any size but are preferably used in large, permanent installations. They can contain the advanced features noted above, are centrally powered, and centrally controlled.
- Distributed systems. These systems can be of any size but are distinguished from centralized systems by the fact that they may be centrally controlled, but the power to speakers is distributed throughout the area.
Masking System Components
All masking systems are composed of several basic components. The primary component is the source of one or more random electrical signals. Typically the source may create either pink or white noise. These particular spectra are seldom acceptable to listeners when converted to masking sound, so a spectrum equalizer is needed to create the proper sound spectrum a listener hears. Most professionals recommend the equalizer cover at least the speech frequency range from 160 Hz to 8000 Hz in 1/3 octave bands; most products cover a broader range. Auxiliary signals, such a paging and music, can be added from an outside source. Those electrical signals are seldom proper when converted to sound, so a more limited spectrum equalizer should be added. All signals are added in a mixer to set their relative levels and then sent to one or more power amplifiers. From the amplifiers, the mixed signal is sent to devices that control the overall level in various areas called zones and thence to either loudspeakers or vibration devices to create sound. Many newer masking systems enclose all components up to the amplifier in one reasonably small cabinet, which can be rack, shelf, or desk mounted. There may be more than one of each component in any system and than one system in a large facility. Most recent system designs have incorporated the ability to control many of the components by software, either locally or remotely. See the dashed lines in the figure. The advantage of such controls lies in the simplicity with which the settings of the various components can be altered. The disadvantage is the added cost to add these controls, especially if system changes are infrequently required.
Masking Speaker Arrays
Offices requiring sound masking, particularly open offices, can have shapes other than rectangular. if the office is large, creating the array manually can be tedious. Several companies have developed programs that can automatically create speaker arrays given the dimensions and shape of the room as well as the desired the vertical location of the speakers (See Section 2.1.3). The program can be used to help develop a materials list and provide a printout for an installer to locate each speaker. An example printout is shown in the figure on the right for an office with a central core of elevators. The vertical and horizontal location of each speaker is shown in the printout as well as the channel, zone, and speaker tap setting. In cases where tap settings vary, the speaker shape is changed.
Horizontal Speaker Spacing and Vertical Locations
The horizontal spacing of masking speakers in commercial facilities has a strong impact on the cost of a project. If the speakers are too close, the cost is higher but the uniformity is good. If the speakers are too far apart, the cost is lower but the uniformity and the acceptability of the masking is compromised.
- Spacing for speakers in a suspended ceiling plenum is determined by:
- Suspended Ceiling Material. A gypsum ceiling permits wider spacing while fiberglass requires closer spacing.
- Suspended Ceiling Height. Higher ceilings permit greater spacing.
- Plenum Depth. Deeper plenums permit wider spacing.
- Sound absorbing batts in the plenum. Batts require Closer spacing.
- For a 3 foot plenum and NRC 0.55 standard mineral tiles, 15 feet is the default spacing.
- Spacing in an open plenum is determined by:
- Structural ceiling Height. Higher ceilings permit greater spacing.
- Sound absorbing batts. Batts require closer spacing.
- Typical spacing for face-down masking speakers is 12 feet to fit into the standard 2 by 4 ceiling tile shape.
- Spacing for under floor speakers is determined by:
- Cavity height: Deeper cavities permit greater spacing.
- Air grilles: Spacing is determined by their presence.
In a Suspended Ceiling plenum
The plenum is the space between a suspended ceiling and the structural deck above it. Since most offices have such spaces, sound masking speakers are commonly added there. For open offices, uniformity of the sound masking in the inhabited area is important and is largely determined by the horizontal spacing of the masking speakers as described above. Speaker arrays are generally rectangular. They generally face upwards to reflect sound from the deck and broaden the distribution of the sound to create a more uniform sound field. The diagram on the right shows an example. For closed offices, uniformity of sound level is not an issue. Generally, ceiling tiles with a high transmission loss are recommended, those with CAC (Ceiling Attenuation Class) ratings of 30 or greater. This reduces the impact of sound masking in adjacent open areas which would have excessive levels in those offices. Typically the ceiling plenum is used as a return air duct requiring an open grill. Sound masking in the plenum above a closed office can generate excessive levels. There are simple metal covers that are placed above the grille to reduce the impact of the hole. Available ones are sufficiently large so do not impede air flow. The upper figure n the left shows a masking speaker that mounts directly on the ceiling tile grid. The two lower figures are examples of speakers that are hung in the ceiling plenum. If the ceiling plenum is shallow, a low profile speaker is required. This speaker radiates sound horizontally rather than vertically to improve the uniformity of the masking in the occupied area below. An example is shown in the lower left figure.
In an Open Ceiling
In some offices, particularly warehouses that have been converted to office space, there is no ceiling plenum. Masking speakers are hung in a similar manner to those above suspended ceilings but the speaker spacing and height is different. Typically, the speakers are mounted higher and the spacing between them is closer. The height of the structural ceiling above is a large factor in spacing, as is the presence or absence of sound absorbing materials on the ceiling surface. Reasonably good uniformity of sound masking can be achieved. Some office designs employ the use of scattered (as opposed to continuous) ceiling tiles (clouds). Careful design is needed to achieve reasonable sound masking uniformity.
Under raised Floors
Sound masking speakers can be placed under raised floors in offices that utilize them. To carry the weight above, the floor material is structurally strong and offers high attenuation of any sound below it. Experience has shown that not only is the uniformity of the sound above exceptionally good, but occupants are hard pressed to determine where the sound source is located (good diffusion -See Section 3.7.6). The figure on the right shows masking sound levels at 48 inches above the floor; they are all within 1 dB of each other. One direction is a line between the speakers; the other direction is a line lateral to the speakers starting at the midpoint. If the depth of the cavity is sufficient, normal sound masking speakers can be used, otherwise small speakers can be used; they fit into a cavity as small as 2 inches and radiate sound horizontally. If the cavity is used as a return air duct, care must be taken to shield the opening.
Face Down in a Suspended Ceiling
Unlike masking speakers above a suspended ceiling or under a raised access floor, the masking sound from face-down speakers travels directly down to a listener without the benefit of any intervening sound attenuating materials. Face-down speakers have been, for many years, typical for paging and music systems, but is now used more recently in some sound masking applications. Initially, the intention was to minimize the visibility of the speaker by having a ceiling-like tile act as a masking speaker. It was first introduced by Bertagni and later commercialized by the Armstrong Company. The picture on the upper right shows a plenum view of this masker. The lower side appears to be a standard fiberglass mineral tile. Because so many devices penetrate ceilings, this speaker's invisibility is appealing to architects. The technical difficulty in developing this product was in broadening the spread of the masking to overcome the lack of intervening material, but the masking spectrum can be adjusted to those typical of those recommended by consultants.. A more recent development was the introduction of a face-down masking speaker much smaller than those used for paging systems. The lower picture shows this masker. Because of small size this masker has a masking spectrum that is not in the range of spectra recommended by consultants. See the recommendations section below. These speakers are often called direct field speakers to distinguish them from the other arrays that radiate sound indirectly through intervening materials. Although the speaker locations in the previous sections are often preferred to increase sound diffusion and acceptability, there are situations where use of direct field speakers is advantageous. An alternative location for the small speakers is mounting on a wall.
Mounted on Panel Tops
Masking speakers have been mounted on the tops of panels for a number of years. Most were associated with Herman Miller furniture systems. Early versions had a spherical ball with an upward facing speaker mounted in it. The ball was placed on top of a short rod which was connected to the top of a furniture panel. The sound radiated upward and reflected from the ceiling tile. Each masker had volume and spectrum controls that were accessible to persons on either side of the panel. Since most open offices had many panels, proper spacing of the masking was not a problem. The effectiveness of the masking was somewhat determined by the sound absorption characteristics of the ceiling tile. This masker was later replaced by the maker shown in the figure on the right; control of the level and spectrum was centralized.
In Unusual Locations
In some older buildings the above locations do not exist. Vibration maskers have been applied in a number of facilities. One application is use on top of a gypsum board ceiling; the entire gypsum panel becomes the speaker. They have also been attached to air supply ducts, radiating masking through the air diffusers. In other cases, maskers were hidden above air ducts in open ceilings. Masking speakers have been placed under desks, under open office work surfaces, and behind paintings. The figure on the right shows a wall mounted masking speaker.
They are a number of situations where a masked area in adjacent to an unmasked area, such as a corridor or waiting area. Experience has shown that abrupt changes in background sound level is noticeable to a person moving from one area to another. Acceptability of the transition is greatly improved by applying soundscaping concepts, where the change in level of background sound is made gradual rather than abrupt. An example is shown in the figure on the right. The area with sound masking is connected to an hallway to other offices. To minimize the change in level from masking to ambient, additional speakers are added to the hallway. The level on succeeding speakers is reduced by 3 dB (a barely noticeable amount) until the ambient level achieved.
Two Design Rules
- The masking should be placed by the listener. Inexperienced persons will often want to place the masking by the talkers. This will require them to speak louder or closer. It will not mask listeners very effectively.
- Every attempt should be made to make the system truly background, conveying no information those exposed to it.
Two Design Objectives
- The masking sound should random and incoherent; meaningful sound should not be used. The sounds from adjacent speakers should not have any correlation with each other. This is best accomplished in two ways: (1) by having intervening materials (e.g. ceilings) between the speakers and listeners; or (2) adjacent speakers are fed a signal from a different masking source.
- The sound field should be as diffuse as possible to reduce awareness. Intervening materials help in this direction.
CHANNELS: The number of places where the masking frequency spectrum can be set. For commercial facilities, one channel is required for each of the different speaker locations noted in Section 2.1.3. For secure facilities, one channel is required for each of the different speaker locations noted in Section 2.3.5.
ZONES: The number of places where the overall masking level can be set (generally in terms of dB(A)). Levels can be set along with the spectrum in some generators. It can also be set at each power amplifier attached to a number of masking speakers. The speakers can be further subdivided into zones with the use of multiple attenuators attached to each amplifier. Finally, each masking speaker should be capable of individual level control. In a sense, each speaker can be made into a separate zone. The primary purpose of such detailed control is to offset the variations of level caused by local geometric influences, such as air ducts. In some advanced systems, zone control can be done centrally.
Common open office levels
Numerous measurements on the privacy afforded by various panel heights have resulted in recommendations for overall masking levels. They are for Normal Privacy with fiberglass ceiling tiles and no obvious flanking paths.
- Panel Height: less than 150 centimetres (59 in) Level: 48 dB(A)
- Panel Height: near 150 centimetres (59 in) Level: 47 dB(A)
- Panel Height: near 170 centimetres (67 in) Level: 46 dB(A)
- Panel Height: near 180 centimetres (71 in) Level: 45 dB(A)
- Panel Height: near 200 centimetres (79 in) Level: 44 dB(A)
Common open office spectrum
The sound masking spectrum shown in the table on the right is one whose overall level is 47 dB(A). This is a default level and can be adjusted based the office design. For higher or lower levels, add or subtract 1 dB at each frequency. See recommendations above. The chosen spectrum contour is based on numerous measurements and is a mean value based on spectra used by numerous consultants. See the upper figure on the left for examples of the range of spectra used. The importance of spectrum contour is shown in the lower figure on the left where a wide variety of masking spectra produce various levels of privacy even at the same overall level.
Common closed office spectrum
Closed offices can be constructed such that Confidential Privacy can be achieved with closed doors and without sound masking. However, the expense to do this is considerable in both materials and in installation labor. More common are STC45 walls that end at a suspended mineral tile ceiling, return air grilles and a continuous ceiling plenum between rooms. It that case, Confidential Privacy can be achieved with sound masking. The table on the right shows a sound masking spectrum that accomplishes that objective under the conditions listed above. The overall level for that spectrum is 44 dB(A). Variations in installation quality have shown that overall levels from 42 to 45 dB(A) may be acceptable.
The Relation of Sound Masking to Acoustical Privacy
The primary goal of sound masking is privacy. In closed offices, conference rooms, or secure facilities the goal is confidentiality of speech. In open areas, the goal is sufficient privacy so occupants are free of distracting sounds, such as speech and other sources of sound. Unfortunately, sound masking is only one of three major factors that result in privacy. Thus, it is necessary to appreciate how the other factors contribute to privacy and how sound masking as a privacy tool can effectively complement those factors.
Three Privacy Factors
The factors are displayed as a diagram in the figure below.
SOUND SOURCE (LEVEL CREATION) Minus SOUND ATTENUATION (LEVEL REDUCTION) Minus BACKGROUND (LEVEL MASKING) = SPEECH INFORMATION
Talker Voice Level
Fortunately, most employees regulate their voices in a responsible manner. Most speak in what is now called a normal voice and this level is well documented in standards publications. At a little over three feet the normal male voice level is about 70 dBA. Most office designs for good acoustics uses this voice level as a criterion for privacy estimation. The loud voice of a male at the same distance is near 76 dBA. Creating good privacy from loud voices, or amplified speech, is almost futile in open offices. It is prudent to consider closed rooms with adequate walls and ceilings for conflict resolution situations. Although women's voices are detectably different than men's, the variations within a gender are more significant than across gender. Use of a male voice is recommended.
Talker Voice Direction
The human voice is very directional especially at the frequencies that are important for intelligibility. The benefit of facing employees away from others is clear; intelligible speech can be reduced by almost 10 to 12 dB. In an open office this magnitude of reduction is difficult to achieve with a furniture system with out great expense. Even at 90 degrees intelligibility is significantly reduced. Since people continually change the direction they face, office design should locate telephones to set a preferred facing direction. The same can be done for positioning visitor chairs.
En route to a listener the level of the sound will be diminished due to several factors:
- Distance. In a location with no reflecting surfaces, the sound will reduce about 6 dB per doubling of distance. While this may sound advantageous at close distances, doubling 25 feet to 50 feet entails considerable area and cost.
- Reflecting Surfaces. Sound reflecting from a surface may be added to the sound going directly to a listener, reducing privacy. Absorptive materials will reduce the level of that reflected sound.
- Blocking Structure. Any material that is in the path of sound from source to listener will reduce the level. The heavier the material, the better. A common error is to use sound absorbing materials to block sound; they are insufficient for this purpose.
- Diffraction. When sound waves encounter the edge of a furniture panel in an open office, the wave bends around the top, side, or bottom. A large bending angle provides more sound loss.
In any environment, there are multiple sound paths with varying degrees of effectiveness in achieving privacy. The weakest path is the one that dominates the potential for poor privacy, which can create difficulties for sound masking.
The background sound in buildings is normally determined by the air handling system. ASHRAE has guidelines on just how loud they systems are recommended to be. The sound from well designed systems is distributed through air diffusers and of low level. It is not capable of being easily adjusted to accommodate privacy needs. The sound spectrum is weighted toward low frequencies so is not useful for improving privacy, especially speech privacy. Sound masking, being electronic, can adjust both level and spectrum at almost any place at any time to meet privacy requirements. However, care must be taken to meet acceptability requirements.
Degrees of Privacy
How people respond at various degrees of privacy is shown in the table.
Talker confidentiality is protected continuously at all times and from all surveillance methods.
Talker confidentiality is protected from casual listeners. Speech may be audible but not intelligible.
Listeners are protected from distracting sounds, especially speech. Speech will be audible, partially intelligible, and not distracting.
Listeners are protected from distracting sounds. Speech will be audible, mostly intelligible, and somewhat distracting.
Listeners are not protected from distracting sounds. Speech will be audible, intelligible, and completely distracting.
Types of Privacy
Listeners or talkers have privacy from all surrounding persons regardless of degree. Well designed closed offices, conference rooms, and secure facilities meet this requirement. Although several degrees of privacy may be applicable, the usual interpretation is Normal Privacy from all surrounding persons
Listeners have privacy from only some of the surrounding persons, not all. Open office customer support centers are facilities that meet this requirement. A person has no privacy from those nearby, but has privacy from those further away. Because this is distance related, a term called Radius of Distraction (RofD) was developed by Thomas Koenig of Dynasound to identify the transition to Normal Privacy (AI<0.2, PI>80). Although the radius is not a perfect circle because of the directivity of the human voice, it does permit the Area of Distraction to be determined. It is the area in which listeners are likely to be distracted by a talker's voice. Since the radius depends on the direction the talker is facing, use of RofD is restricted here to the direction the talker is facing (worst case). Knowing approximate workstation size, the number of distracted persons can be estimated from this area. In terms of A-weighted levels, a talker facing a listener is about 10 dBA louder than when he or she is facing away. Since speech intelligibility is of concern for the Radius of Distraction, voice directivity can be used to calculate it. The distance at which the Privacy Index reaches 80 with a male normal voice, a nine foot high, NRC 0.55, ceiling, and a sound masking level of 46 dBA was calculated and is shown in the figure on the right. Each circle is a five-foot increment. The radius in the talker's direction is about 38 feet while behind is only 10 feet. The considerably smaller area is that for a male casual voice. The RofD and Area of Distraction for several voice levels are shown in the table on the left. It is clear that voice level is the critical factor in Partial Privacy situations where there are no separating panels of acoustically significant height. Without sound masking, the RofD is on the order of 41 feet, covering about 2250 square feet. The beneficial effect of sound masking level on RofD is shown in the figure on the right. The value of the improvement can be better appreciated by estimating the Area of Distraction with the formula: 1.3 times the square of the radius.
The Privacy Index
Since speech privacy is the most important sound that needs evaluation, The Privacy Index has been developed; it based on the well established Articulation Index. The privacy calculation is done for each of several frequencies important for speech. Each frequency is then weighted for its importance in intelligibility and summed up. Thus it is a measureable means to ascertain the ability to understand speech and thus determine speech privacy. Numerous tests over fifty years have shown a strong relationship of that metric to the degrees of privacy. The value of PI varies from 0 (no privacy) to 100 (complete confidentiality). The curve in the figure shows that the relationship between speech privacy and Privacy Index is not a straight line, suggesting that the brain goes from nearly complete understanding of speech to virtually none with small changes in the PI. Sound masking applications make use of this fact. If the other two factors can move the Privacy Index to near 70 (the knee of the curve), addition of sound masking will have a large effect on the degree of privacy despite the fact that the PI does not change significantly.
Although people tend to think most relationships are linear (halfway in one direction yields halfway in the other), they are very familiar with non-linear school grades (50 in an exam is flunking). This analogy can be used as below:
- Secret PI=100+ Grade=A+
- Confidential PI=95–100 Grade=A
- Normal PI=80–95 Grade=B
- Transitional PI=60–80 Grade=C
- None PI<60 Grade=F
Other Privacy Measures
A newer rating is the Speech Privacy Class (ASTM E2638-08); it applies to closed rooms. PI is a measure of the speech from a talker at a specific location to a listener at a specific position. It is applicable to situations such as open offices as well useful for evaluating the efficacy of a listening device hidden in a stud wall. SPC makes use of average values so is very useful in the design and evaluation of closed rooms for speech privacy from casual listeners. There are several other measures such as, Speech Privacy Potential (SPP), Speech Intelligibility Index (SII), Speech Transmission Index (STI), Speech Interference Level (SIL) and Speech Intelligibility Contrast (SIC). Discussion and definitions of these ratings can be found in the literature. The advantage of the Privacy Index is that comparison of that rating to school grades makes the designer and user feel more comfortable with the rating; some of the others are more technical.
Why Calculate the Privacy Index?
In offices, it is seldom practical to make detailed measurements of the privacy provided by a sound masking system. The vast number of acoustical interactions in even small offices makes it a long and time-consuming project. As specialists gain more experience, setting up and adjusting the system to increase user acceptability will become just as important as detailed measurements. However, there are circumstances where such measurements are valuable:
- If a specification calls for the installer to take such measurements.
- If an existing office looks marginal for creating privacy. Reporting such is important.
- If facility managers need objective data to offset unreasonable complaints or justify expense by showing the improvement provided by masking.
- If the privacy achieved needs to be proven or in secure facilities at locations where listening tests cannot be performed (e.g. windows).
Privacy Index can only be calculated from a series of measurements, since it incorporates all relevant factors for speech privacy. For normal masking systems, the acoustical version is used. For secure facilities, where speech may be in vibration form, the second method must be used. To collect the necessary data, a one-third octave band Real Time Analyzer with a random incidence microphone is necessary. To create the test sound, a source with directional characteristics similar to that of the human voice is necessary. A tripod should be used to mount the sound source which is normally 48 inches high for seated talkers.
The Acoustical Privacy Index
The equations for Articulation Index and Privacy Index are:
The transmission loss spectrum (TLi) must be determined from two measurements; a total of four spectrum measurements are indicated. It is both common and acceptable to choose the normal voice spectrum (VSi) available from ASTM. The source spectrum (SSi) should be measured independently, since it can be used for multiple PI measurements. The received spectrum (RSi) and masking spectrum (MSi) must be measured at the listener location, but not at the same time. Since there are many combinations of talker/listener positions even in one workstation pair, it is best to choose a worst-case situation such as shown in the figure. The sound source (dark symbol) should face the upper workstation at seated height. Although the talker may seldom be in this position, it presents the worst privacy situation for the other parties. The listener should be at the most frequently used position; in this case, at a workstation near the talker (open chair symbols). For a second test, placing the sound source (dark symbol) at the workstation table, and facing across it, would be the worst-case for the listener in the workstation to the right.
- Step 1: The source spectrum (SSi). The sound source, mounted on a tripod, should be faced horizontally at seated height (48 inches) and pointed in the desired horizontal direction. It should be powered by a broadband random generator/amplifier. The spectrum should be as near pink (flat one-third octave band spectrum) as possible with an overall level of between 70 and 80 dBA. However, since it will be used as part of a level difference, it need not be exactly pink. It is preferable to take this measurement in an anechoic environment but such rooms are hard to find. Further, one may have to determine the spectrum in the field. In that case, the source should be at least fifteen feet from all sound reflective vertical surfaces and placed over a carpeted floor. To minimize the destructive interference of the floor reflection, measurements should be made at 2, 3, and 4 foot distances along the speaker axis. Arithmetically averaging the three spectra tends to strongly reduce the interference, which is below the frequencies important for speech anyway. Using software or Excel, the averaged spectrum can be stored and retrieved so source spectra need only be entered once.
- Step 2: The received spectrum (RSi). With the sound source on and the sound masking off (to avoid interference), the microphone should be placed at the listener’s position, 48 inches high. To avoid body reflections, the microphone may be placed on another tripod, or held at arm’s length at an angle not along the sound path.
- Step 3: The masking spectrum (MS). With the sound source off and the sound masking on, make a measurement at the same received position.
- Step 4: Calculate Privacy Index wit available software.
To minimize the tedium of these measurements, the source can be set up in one workstation, then rotated to make received measurements in all surrounding workstations. In that case, four or five measurements can be made relatively quickly.
The Vibration Privacy Index
All spectra for the Acoustical Privacy Index calculation are acoustical; not so here, both acoustical and vibration spectra must be measured. In this case the received spectrum is the vibrational response to the acoustical sound source (VRSi)and the masking spectrum is the vibrational response caused by the vibration masker (VMSi). To measure these, a vibration detector must be used. Because the equation requires only the difference between the two spectra, calibration of the detector is not necessary. It only requires sufficient gain to detect all relevant spectra. The equations are:
These equations should be used for windows and in other circumstances where surveillance vibration detectors are likely to be used. The normal speech spectrum is often used (VSi).
- Step 1: The source spectrum (SSi). The same sound source spectrum as used for the Acoustical Privacy Index may be used here. The overall level should be at least 80 dB(A).
- Step 2: The received spectrum (VRS). With the sound source on and the sound masking off, the vibration detector should be placed at the listener position on the surface. Make a measurement of the received vibration spectrum.
- Step 3: The masking spectrum (VMS). With the sound source off and the sound masking on, make a vibration measurement of the vibration masking spectrum at the same position on the surface as for the received spectrum.
- Step 4: Calculate Privacy Index with appropriate software.
Achieving Acoustical Privacy with Sound Masking
Ratings for Sound Reduction
Detailed descriptions and definitions of the various sound product ratings can be found in standards and in the literature. Here their relevance only to speech privacy is discussed.
These ratings describe how much sound is NOT reflected from a surface, such as ceiling or furniture panels. They vary from 0 (complete reflection) to 1 (no reflection). The higher the rating, the less the reflected sound. They do NOT express the reduction in sound level and do NOT relate specifically to speech. Noise Reduction Coefficient (NRC) is not necessarily a measure of how much sound is converted to heat. It is actually a measure of how much sound is NOT reflected. Misinterpretation of this difference has resulted in furniture panels that are completely fiberglass with extremely high NRC ratings. Unfortunately, speech is not absorbed but passes through the panel to the person in the next workstation, with disastrous privacy. Sound Absorption Average (SAA) is an updated version of NRC.
These ratings evaluate the loss of sound passing through a material. In each case the higher the rating the more sound loss. The number can loosely be related to the loss of sound level and do NOT specifically relate to speech. Sound Transmission Class (STC) is a standardized rating typically used for various partitions such as walls or furniture panels. Noise Isolation Class (NIC) is a simpler rating similar to STC. Ceiling Attenuation Class CAC) is a rating similar to STC except that it is specifically designed to handle sound transmitted between two closed offices through a ceiling plenum.
Sound Bending (Diffraction)
When light waves, water waves, or sound waves encounter the edge of an object, the wave bends around the edge; it is called diffraction. There is no rating for this phenomenon but it can be calculated when necessary. It is most important with furniture panels in the open office. The line from a talker to the panel edge and then to a listener creates an angle. The greater the angle the more sound reduction. A large bending angle occurs when the talker or the listener is closer to the panel and when the panel is high.
These ratings are directly related to the loss of sound level making them advantageous for evaluating product performance in offices. They have a further advantage in that they relate specifically to speech. The loss of sound at any frequency is weighted by its contribution to speech intelligibility at that frequency. They can be used to evaluate product performance in terms of speech, but since privacy is determined by the combination of sound multiple paths, they cannot evaluate the overall performance of a particular situation. Articulation Class (AC) is a standardized rating for speech loss. Speech Loss (SL) is calculated the same as AC but the value is one-tenth that of AC so is more closely comparable to the loss of intelligible speech. Effective Speech Loss (ESL) is similar to SL except that it also takes into account the presence of sound masking. Essentially it combines the intelligibility loss due to the structural loss with the loss caused by masking. It helps to evaluate the contribution of sound masking to speech privacy. The equations for these are given below.
TL is the transmission loss in decibels, MS and BS are respectively masking and ambient sound levels in decibels. The speech weighting factors W are those listed with Articulation Index. The first two ratings account for the high sound loss at frequencies that are important for intelligibility. ESL treats the difference between the masking spectrum and the background spectrum as an additional loss. The figure on the right shows the Speech Loss for a furniture panel with high STC; it is the excess loss caused by sound diffraction over the panel. The Effective Speech Loss is higher and shows sound masking as if it were part of the partition. The masking level for this example was set at 47 dB(A) and the background level was 39 dB(A). This graph demonstrates the significant value of sound masking when it acts as an additional panel height. It is clear that the level used in the graph only would be necessary for low panel heights, but is excessive with higher panels. The graph also demonstrates the panel tradeoff with sound masking. Adding masking permits lower panels to achieve the same speech loss. Adding masking roughly permits a reduction of at least 12 inches in panel height so there can be a significant cost savings.
The reflection of speech from a ceiling in an open office depends on the sound absorbing characteristics of the ceiling tile and the height of the suspended ceiling. Because ceiling tiles tend to absorb the higher frequencies better, the speech loss is relatively good, as seen in the figure on the right. It should be noted that an upgrade in NRC rating adds expense but not always appreciable speech privacy improvement. A goal of SL=20 for the ceiling would require the panel height to be at least 60-66 inches to balance the two paths with added masking, otherwise, the panel would be the weakest sound path. Higher ceilings also increase sound diffusion for masking speakers in the ceiling plenum, increasing acceptability.
The recommended open office masking levels in the section above are based on the speech loss of all sound paths in the particular situation. The weakest path, normally, is the one that controls the amount of privacy afforded. The speech loss of panels and ceilings are major contributors in open office workstations, but not all. The figure on the right shows these levels for a free workstation, one that has no other surrounding surfaces, such as windows or walls. The levels are based on the workstation occupant having Universal Normal Privacy (freedom from distractions by all surrounding persons). This is the most restrictive requirement but despite this, 60-66 inch surrounding panels work acceptably.
Modeling Speech Privacy
The advantage of modeling is that an analysis can be made before an office design is installed. The disadvantage is that speech privacy is between two persons and each open office has hundreds of such relationships. Often this disadvantage can be minimized when many of the workstations are similar. Critical workstation designs should be analyzed separately. Any modeling program is built on acoustical equations and then tested in the field. The program must take into account various office arrangements (closed and open), various sound attenuation paths, and various masking spectra and levels. Several companies have developed such modeling programs. The purpose of the modeling is to determine if the speech privacy desired by the owner can be achieved with reasonable levels of sound masking. If not, the model should be able to show how the office arrangement can be improved or what speech privacy to expect at reasonable levels of masking if not improved. All too often, the masking system installer is expected to provide the desired privacy in circumstances that require excessive levels. The discussion below is derived from a specific program that can be downloaded free from the internet.
Choosing an Open Office Workstation Design
The first task is to choose the various elements that pertain to acoustical privacy. There are three materials that need to be chosen: the ceiling, the panels or walls, and the carpeting. The program should contain a data base of acoustical characteristics for each of these materials. The figure on the right shows a typical modeling screen for an office design. The chosen ceiling had an NRC of 0.98 (fiberglass), the panels had an STC of 24, an NRC of 0.75 and all were 66 inches high. The carpet was wear resistant and the ceiling height was chosen to be a common 9 feet. The workstation size and the occupant positions are as shown in the figure. The occupants could be made to face in various directions. No wall or window reflections were present in this design, nor did the workstation have panel shelving. It was possible to add one of three light fixture types at any location in the ceiling. This program had a Quick Analysis button that gave the final degree of privacy and the required masking level quickly with no details; the purpose being to detect poor designs before detailed analysis was made.
Calculating Sound Attenuation
Once the workstation design has been defined, the sound attenuation from the talker to the listener needs to be determined. An example of this calculation is shown in the figure on the right. The workstation design is the same as the one in the previous section. Because of the wide variety of designs the program contains forty-six possible sound paths, each path can be examined independently (Analyze One Path) or collectively (Analyze All Paths). they include direct paths, first reflections, and second reflections. Although direct and first reflection paths normally determine the amount of sound attenuation, on occasion lesser paths contribute to the degradation of privacy. Although the time delays of each path are different, the differences are too short to be noticeable by the listener. The direct path only is shown in the figure, speech through the panel and speech over the panel. All calculations are done in 1/3 octave bands and the resulting sound attenuation spectrum is shown on the right of the figure. The various rating results are shown at the bottom and are ranked in the table from least loss to greatest loss. If the loss is greater than 60 dB, it is not displayed. When Analyze All Paths is chosen all paths are summed and the overall loss is displayed. The performance of the various furniture and ceiling systems in the database can be compared; a table of the various sound attenuation results will be displayed.
Adding Sound Masking
Once the sound attenuation characteristics of the workstation pair are determined, it is necessary to determine whether sound masking is needed, and if so, how much. The figure on the right shows the final result for the design listed in the sections above. There are several variables. Voice level is first. Most people in offices speak at conversational (normal) levels. Raised levels can be used for speaker phone use. The gender of the speaker has only a small effect on intelligibility. The background level in the space is second. There are three choices. The first is a level and spectrum that is typical of well designed air handling systems. The second is the Walkaway Test. In an open area, a person reads text at normal voice levels and another backs away until he or she has difficulty understanding the text. The ceiling tile array can be used to determine the distance. That distance is entered and the typical background spectrum level is altered based on tests that had been made. The third is actual measurement. The level in dB(A) is entered and the typical background spectrum level is altered. The spectrum shape (not level) for the sound masking must be chosen. Choosing fixed opens a list of masking spectra used by various consultants and installers. from which the user must pick. The first two choices are the spectra shapes recommended in the sections above. A choice of Automatic allows the program to create a spectrum shape that attempts maximize privacy with minimum level. The privacy goal must then be entered as shown in the upper right. A choice of Confidential Privacy defaults to a Privacy Index (PI) of 95. A choice of Normal Privacy defaults to PI=80. The third choice permits any Privacy Index to be chosen as a goal. There are cases where the sound masking level is specified, so the fourth choice can be taken. At this point there are options for setting the masking level. The first is Tune; the chosen masking spectrum is visually adjusted in level until the chosen degree of privacy is obtained or the specified masking level is reached. Then the various privacy, level, and sound attenuation ratings are displayed. A second choice is available that performs the same task with each of the masking spectra in the database and displays the comparative results in a table.
Some Results of Modeling Open Offices
Normal speech privacy is very difficult to achieve in corner workstations with panel heights between 60 and 66 inches and with the masking levels recommended above. Workstations along a wall have reduced privacy due to the wall and the wall-ceiling reflection. For workstations away from major reflecting surfaces the table on the right gives some guidance for what can be achieved with the three key factors of panel STC, ceiling NRC, and sound masking level. A high STC rating of a panel is not needed for low panels, the weakest path is the sound over it. There are three general classes of ceilings. The more common and less expensive ones have ratings from 0.55 to 0.65. The high performance ones have ratings between 0.65 and 0.85, and the more expensive fiberglass tiles have higher ratings. The table shows the estimated Privacy Index for combinations these materials along with appropriate sound masking levels. Masking levels above 47 are not generally recommended
- Architectural acoustics
- Cone of Silence
- Noise mitigation
- Noise Reduction Coefficient
- Noise regulation
- White noise machine
- Tinnitus masker
- Active noise control
- Kryter, K.D (1970), The Effects of Noise on People, New York: Academic Press.
- Evans, G.W; Johnson, D (2000), "Stress and Open Office Noise", Jo. of Applied Psychology, 85 (5): 779–83.
- Banbury, S; Berry, D.C (1997), "Habituation and Dishabituation to Speech and Office Noise", Jo. of Experimental Psychology: Applied, 3 (3): 181–95.
- Banbury, S; Berry, D.C (1998), "Disruption of Office Related Tasks by Speech and Office Noise", British Jo. of Psychology, 89: 499–517.
- Jones, Dylan. "The Cognitive Psychology of Auditory Distraction", British Jo. of Psychology, 90, 167-187, 1999.
- Chanaud, R.C., Concepts of Masking Sound, So. & Video Cont., Feb 1986
- Chanaud, R.C., Sound Masking Systems, Sound & Comm. 60-64, Aug 1984
- Sound Masking, Best noise machine, retrieved 2014-09-16.
- Chanaud, R.C., "Sound Masking Done Right", Magnum Publishing, 2009
- Anon. "Health Insurance Portability and Accountability Act", 45 CFR, Parts 160 and 164, 2002
- Anon. "Protecting the Privacy of Patients Health Information", U.S. Department of Health and Human Services", Fact Sheet, April 14, 2003
- Anon. "Oral Communications", U.S. Office of Civil Rights, Fact Sheet, April 5, 2002
- Anon. "Can You Hear Me?", New England Journal Of Medicine, 289, 803-804 Oct 11, 1973
- Anon. "Noise in Hospitals", U.S. Department of Health, Education and Welfare, Public Health Service, 1963
- Anon. "Unquiet Rest", British Medical Journal, 715-716, Sept. 23, 1972
- Falk, S.A. and Woods, N.F. "Hospital Noise Levels and Potential Health Hazards", New England Journal of Medicine, 774-780, Oct 11, 1973
- Gedeke, R., Doring, F, Keller, F. and Vogel, A. "The Noise Level in Children’s Hospital and the Wake-up Threshold in Infants", Acta Paediat Scand., 58, 164-170, 1969
- Haslam, P. "Noise in Hospitals: Its Effect on the Patient", Nursing Clinics of North America, 5, 715-724 Dec. 1970
- Hurst, T.W. "Is Noise Important in Hospitals?", Journal of Nursing Studies, 3, 125-135, 1966
- Statham, C. "Noise and the Patient in the Hospital", British Medical Journal, 1247-1248. Dec 5, 1959
- Welch, B.L. and Welch, A.S. "Physiological Effects of Noise", Plenum Press, New York, 1970
- Topf, M. "Stress Effects of Personal Control over Hospital Noise", Behav. Med. 18(2), 84-94, 1992
- Anon. "Information on Levels of Environmental Noise Requisite to Protect Public Health and Welfare with an Adequate Margin of Safety", Document 350/9-74-804, U.S. Government Printing Office, 1974
- Suter, A.H. "Noise and Its Effects", Administrative Conference of the United States, Nov. 1991
- Sanders, J. "Quiet on the Hall: Researchers Search for Ways to Reduce Noise and Improve Sleep in Nursing Homes", Georgia Institute of Technology Research News, Oct. 2002
- Anon. "Physical Security Standards for Sensitive Compartmented Information Facilities", Defense Intelligence Agency Manual 50-3, 1980
- Anon. "Physical Security Standards for Sensitive Compartmented Information Facilities", Director of Central Intelligence Directive 6/9, November 2002
- Anon. "Construction of Secure Conference Rooms", Air Force Pamphlet 88-26, July 1990
- Anon, "JAFAN 6/9, Physical Security Standards, Special Access Program Facilities", Joint Document of the US Army, Navy, and Air Force, March 2004
- Anon. "Gramm-Leach-Bliley Act," 15 USC SubChapter1, Sections 6801-6809, 1999
- Chanaud, R.C., Active and Passive Audio Surveillance Countermeasures, Guest Speech at International Security Society Conference, Washington, DC 1991
- Chanaud, R.C., Building Design Crucial to Audio Security, Access Control, V 36, No 1, 32-38, Jan 1993
- Chanaud, R.C. "Progress in Sound Masking", Acoustics Today, October 2007
- Chanaud, R. "Sound Masking Done Right", Mitek Publishing, 2009
- Chanaud, R. "Sound Masking Done Right", Mitek Publishing, 2009
- Chanaud, R. "Sound Masking Done right, Mitek Publishing, 2009
- "Test Method for Objective Measurement of the Speech Privacy Provided by a Closed Room". Committee E33. doi:10.1520/e2638-08.
- "Explaining the Speech Privacy Potential Rating" (PDF). Cambridge Sound. Retrieved 17 June 2016.
- "SII: Speech Intelligibility Index". Acoustical Society of America (ASA). Retrieved 17 June 2016.
- "Improving speech privacy in personal sound zones". Xplore. IEEE. doi:10.1109/ICASSP.2016.7471687.
- Kryter, K.D., The Effects of Noise on Man, Academic Press, 1970
- Chanaud, R.C., Sound Masking Done Right, Mitek Corporation, 2009