October 2006
Cheat Sheet: Designing A Sound Masking System
To enhance speech privacy, successful implementation of sound masking systems requires knowledge of acoustics, psychoacoustics, and proper sound system design.
The ABCs of sound masking:
Absorb
Block
Cover
WHAT IS SOUND MASKING?
A sound masking system emits low-level, non-distracting masking noise designed to reduce speech intelligibility and improve speech privacy. A typical sound masking system consists of a masking noise generator, an equalizer, one or more power amplifiers, and a group of special loudspeakers installed above a dropped ceiling.
THREE STEPS TO SUCCESSFUL SOUND MASKING
1. Attenuate the direct sound.
Direct sound from a talker reaches a listener by the shortest path, without being reflected by any object.
2. Reduce sound reflections.
Reflected sound from a talker reaches a listener after being reflected from one or more hard objects.
3. Raise the ambient sound level using sound masking.
Sound masking adds low-level background noise to reduce the speech-to-noise ratio and intelligibility.
INSTALLATION TIPS
1. In a typical ceiling, mount the masking loudspeakers low in the plenum space with the bottom of each loudspeaker about 6 to 8 inches above the ceiling tile.
2. Point the loudspeakers upward at the hard deck.
3. Space the loudspeakers about 12 to 14 feet apart horizontally. The sound will reflect off the deck down through the ceiling tile and into the space below.
COVERAGE
Walk the space, and listen and observe the meter to find any “hot spots.” Determine the cause of any hot spots (open return air grille, etc.), and fix the problems before re-adjusting the masking system (install return air boot).
• Remember to check levels in any walled offices served by the masking system. If the level is too high in these walled offices, which is typical, adjust the transformer taps for the appropriate masking loudspeakers. If the system has zone controls, adjust them to achieve the same level in all zones.
• Re-check all areas, adjusting loudspeaker transformer wattage taps as required. If uneven coverage remains, re-orient, move, or turn off loudspeakers as required.
ASTM E 1130
The American Society for Testing and Materials (ASTM) has developed a standard test method for the objective measurement of speech privacy in open offices using the Articulation Index, which describes the sound source and method of qualifying it using a special qualification signal.
DEFINITIONS
Articulation class (AC) — The sum of the weighted sound attenuation values in the one third octave bands between from 200 Hz to 5,000 Hz. For sound masking in open-plan offices, design for an articulation class greater than 200.
Articulation index (AI) — A number ranging form 0.00 to 1.00, which is a measure of speech intelligibility. The higher the number, the greater the intelligibility.
Noise criterion (NC) curves — A series of curves of octave band sound spectra used for rating the “noisiness” of an occupied indoor space. The actual noise spectrum is compared to the NC curves to determine the NC level of the space.
Noise isolation class (NIC) — A single-number rating derived from measured values of noise reduction between two enclosed spaces connected by one or more paths; this rating isn’t adjusted or normalized to a standard reverberation time.
Noise reduction coefficient (NRC) — A single-number rating of the sound absorption properties of a material such as acoustical ceiling tile or acoustical wall panels; the arithmetic mean of the sound absorption coefficients at 250, 500, 1,000, and 2,000 Hz, rounded to the nearest multiple of 0.05. A value of 0.90 or higher is preferred, with 0.70 minimum value for adequate speech privacy.
Pink noise — The most common masking noise, it’s an electronically-generated, random noise signal that has equal energy per octave or fractional octave bands. Pink noise appears as a flat response on third-octave analyzers.
Room criterion (RC) curves — A series of curves of octave band spectra used for rating the noisiness of an unoccupied indoor space. The actual noise spectrum is compared to the RC curves to determine the RC level for the space.
Sound absorption coefficient — The fraction of the randomly incident sound power absorbed (or otherwise not reflected) by a material. Sound absorption coefficients are given in octave bands for many materials with values ranging from 0.01 (very hard surface such as polished marble) to 1.00 and higher (very absorptive products such as glass fiber boards).
Sound transmission class (STC) — A single number rating used to compare the sound insulation (isolation) properties of walls, floors, ceilings, windows, or doors.
White noise — An electronically-generated, (usually) random noise signal that has equal energy per equal bandwidth. White noise appears as a rising response (3 dB per octave) on octave or third-octave analyzers.
Tools
• Sound level meter
• Measurement microphone
• Spectrum analyzer
• Oscilloscope
Procedure
1. Set all amplifier input controls fully counterclockwise, and set the system equalizer controls to the flat position (unity gain).
2. Place the measuring microphone in a typical listening position at ear level.
3. Increase the gain for the amplifier feeding this listening position to 50 dB SPL in the 500 Hz, 1/3-octave band.
4. Ensure the amplifier isn’t clipping. Connect the oscilloscope to the amplifier output and look for squared-off wave tops.
5. Increase the amplifier gain 10 dB to 60 dB SPL at 500 Hz, and check for clipping.
6. Adjust the equalizer to the desired curve between 200 Hz and 5,000 Hz.
7. If it seems necessary to adjust any individual equalizer control more than 3 dB above unity gain, try moving the mic.
8. If low frequencies can’t be adjusted to this elevated level, re-adjust these bands later when the sound masking level has been reduced to the final level.
9. Fully reset the amplifier gain counterclockwise.
10. Repeat the previous steps for each channel in the system, ending with all amplifier controls fully counterclockwise.
11. Finally, equally increase the amplifier gain controls on all channels until the sound level meter reads 47 dB(A) at all listening positions. If necessary, readjust the low-frequency bands.
12. Fine-tune the equalizer so that the building HVAC noise and masking sound together achieve the desired spectrum.
13. Check the coverage at additional listening positions.