People overhear conversations that weren’t meant for them. This creates problems ranging from innocent distraction to more serious breaches of confidentiality. Assuming you can’t acoustically isolate people, the solution is additive sound masking. Traditionally, our industry throws in some background noise. Problem solved, right? Not so fast. There are many issues to consider in creating an effective masking system. Indeed, there are multiple applications for masking systems that will lead to different design principles.
The primary use for these systems is to reduce unwanted distraction. Research from the University of California, Irvine, indicates that the average office worker is interrupted every 11 minutes, and it can take 23 minutes to regain prior level of focus. Systems meant to address the problem of distractions typically are referred to as energy masking systems.
This issue of distraction brings us to our first application of a masking system, which is to reduce the effect of unwanted sounds in a focus environment such as an office or place of study. In this case, the goal is to decrease distraction by introducing consistent and controlled noise that will reduce the intensity of distracting speech or noises. Imagine trying to carry on a conversation in a crowded hallway or busy street; not always an easy task because adding sound makes speech less intelligible. As an added benefit, when you can’t understand what someone is saying, his or her words become less distracting.
This type of energy masking increases the ambient room noise energy level to a point where distractions below a certain level are less likely to cause work interruptions. This helps cover up distractions such as coworker phone conversations and adjacent cubicle conversations. The key here is to measure the average working noise level of the room and create a system that is 3dB to 5dB louder than the occupied noise level of the work environment.
For systems addressing distraction complaints, it is important that the solution itself doesn’t become a distraction. To this end, the common practice is to set up a shaped white noise signal that is meant to emulate a building’s air conditioning system. Although it can be effective in a fully populated office environment, as the building occupancy drops, the masking system can become obvious and unnatural, especially in a smaller office setting where the level of AC noise is not consistent with the volume of the office space.
We can overcome some of the issues of a “shaped white noise”-only system by taking advantage of two techniques.
First, when ambient noise level patterns are predictable, a DSP with a scheduler can be used to follow typical workday patterns. Figure 1 shows a level diagram with a fade-in at the start of the day, a dip at lunch and then a fade-out at the end of workday.
Alternatively or in tandem, microphones can be used to estimate the noise level of the room and adjust the masking noise levels. Fine tuning requires some intelligence because the system will have to subtract the level of the added masking noise from the ambient sound sample. Generally speaking, the scheduling approach is the method with the greatest ratio of success.
Besides adjusting the volume of the masking system, we can introduce other environmental sounds that could be accepted as a natural part of the environment. A foundation of air conditioning sound could combine with low-level consistent conversation noises or office equipment sounds, such as copiers and printers. Because this technique uses a range of noises, it is more difficult for the subconscious to filter out and the system is more effective at the same SPL level.
In practice, the best distraction sounds are not completely uniform. Some examples include street sounds, running water or unintelligible conversation. These signals are constant enough to drown out distraction without being so consistent that they can be filtered out by our listening minds.
Another element is to protect security of sensitive information. Speech privacy systems often add masking energy outside of the areas, such as a conference room or an office, where conversations are taking place. These are commonly used in legal, corporate and medical offices, as well as government buildings, for example.
Privacy systems rely less on their ability to be discreet and more on their ability to provide a private meeting space. The system should deliver a consistent level of privacy protection when the system is engaged. If the required noise level is intrusive, users might turn it on and off, as privacy is desired.
Although shaped noise systems are often used as privacy systems, they may not be effective at low volumes. Human speech differs quite significantly from a wider spectrum-shaped white noise, and the human brain is quite good at filtering extraneous sounds when we want to focus on a particular source. Simply put, we can hear what we are listening for.
For privacy situations, we want to use sounds that are closer in frequency and pattern to human speech to mask the intelligibility of private conversations. As mentioned earlier, examples include crowded streets or running water, which are much more effective and difficult for the human mind to filter. This means that they can be used at lower volume levels to create a more effective privacy system.
Using synthesized human speech sounds helps eliminate the ability to focus on speech frequencies that enable eavesdropping. And, for cases where the privacy system is located in an area where a chatter-type of sound may be unnatural or disturbing, the base of chatter can be overlaid with environmental sounds, such as that babbling brook or background music.
Figure 2 shows the improved efficiency of a synthesized speech sound over traditional shaped-noise systems. You can read this two ways: Synthesized speech provides a higher level of privacy than shaped noise at the same SPL, or can be used at a lower SPL for an equivalent amount of privacy.
This brings us to the question, “Why not just play music outside of private meeting rooms where eavesdropping is a concern?” Although music can aid in the application of a privacy system, music is dynamic in nature; it rises and falls in level. As a result, when the music becomes quiet, conversations can be overheard. Therefore, music itself does not guarantee privacy, which is why we use other base sounds that provide a minimum level of privacy at all times.
Privacy systems may have to address listening spaces beyond the hallway outside the door. For instance, air ducts to adjacent spaces may require treatment. Or, if more sophisticated active listening is a concern, the system may counter active-listening devices in secure plenum spaces or even laser listening devices on exposed windows. All of this depends on the client and intended use.
Speech privacy techniques have evolved over the years, with growing needs in open offices, medical and government institutions. With careful planning and use of simple tools in modern audio processors, it is possible to create systems that are a perfect fit for any situation.
We hope these suggestions will provide inspiration to try new techniques for masking systems. In the end, our efficiency is measured by our customers as they experience an increase in productivity or improved privacy. As the saying goes, “There is more than one way to skin a masking system!”
In many masking systems, speakers are placed in plenum areas facing away from the room. This helps reduce the issue of localization of the speakers and camouflages the added noise, but it limits the use of the masking-system electronics to a single function and reduces the frequency response of the system in key ranges.
Some systems have found innovative ways to combat these issues while maintaining the ability to use the system for multiple purposes. To reduce localization of the masking noise, the speakers are wired with two channels in alternating patterns: The two channel-masking sounds are delayed from each other at between 11 and 30 seconds. This technique proves quite effective, and because the delay is built into the masking sound generator, it doesn’t affect the paging and BGM sources.
Staggering the ambient noise soundtrack reduces the probability of comb filtering and null spots in the audio that can be distracting.
This approach allows the same DSP, amps and speakers to double as music and paging systems. The masking system is also enhanced with greater frequency response and consistency across all zones, allowing a greater range of sounds to be used to reduce word recognition in a privacy system.
Links related to SPS: