Speech Privacy Class or SPC is a new ASTM
standard of measurement designed to measure and
precisely calculate how hard it would be to
eavesdrop and understand speech from outside of
a room. SPC is an improvement over previous
speech privacy standards such as Articulation Index
(AI) in a number of ways. One major improvement
is that the investigator actively chooses weak points
in the construction such as windows and doors as
test locations. Starting in August of 2004, John S.
Bradley and Bradford N. Gover of the CNRC
published a new index called the “Speech Privacy
Index” or SPI which was then revised to Speech
Privacy Class. This was in part due to State of the
Art Acoustik’s prior work done in testing and
designing new solutions for the Justice Building on Parliament Hill that determined that the testing
and rating method in use at the time to determine privacy was inadequate and needed improvement.
SPC is much more useful at determining speech privacy or speech security than Sound
Transmission Class (STC) or Noise Reduction (NR) because it is intended to measure intentional as
opposed to casual eavesdropping as explained below. Speech Privacy Class depends as much on
the sound isolation performance of the partition as it does on the background noise, since it is
effectively a measure of the signal to noise ratio performance of the space. This is totally unlike an
STC or NR measurement, as these do not take into consideration the noise in the listener’s sound
field. No matter what the background noise level, one should always get the same value for STC or
NR. However any change in the noise will change the Speech Privacy Class rating, since any
change in the background noise level effectively changes the amount of masking of the received
speech. The higher the noise, the more difficult it is to understand speech. The better the partition
performance, the less speech is transmitted. The SPC rating combines these two effects that are
fundamental to speech intelligibility, or in this case, lack thereof.
Older speech intelligibility standards were inadequate for a number of reasons. They assumed the
sound field in the source and receive room were acoustically diffuse; meaning that the sound
generated or received is equally loud everywhere in a room. This may be partly true for the middle
area of common rooms, but not in places where someone typically might try to listen, for example
close to a door in a hallway, or a glazed wall panel or at a window mullion. Then even when one had
obtained a transmission loss at a given location, one applied Articulation Index measures, but this
index was designed to measure how good intelligibility is, not how bad it is, which is what is required
for speech privacy. SPC was laid out to try to overcome these disadvantages.
The test preferably uses omni-directional sound sources, (speakers that play equally loud in all
directions) placed at two points around the room. This greatly helps to simulate an acoustically
diffuse sound field in the source room which reduces error in the measurement, representing
multiple potential talker locations. Instead of a random selection of receive test points in the
receive room such as those used for an STC test which must be at least 1m from any surface, weak
points are subjectively listened for with the source turned on,
Table 1: Above is a classification table describing various SPC values. The table displays the category of speech
privacy or security and the corresponding SPC value needed to achieve the desired level.
and microphone placement is chosen based on actual or expected trouble spots. For example the
microphone might be placed close to the side or even bottom seal of a door. All microphone points
are made at 0.25 meters (10 inches) away from a boundary, and the received level is measured for
each source position at every receive location. The SPC value is obtained by measuring the partition
performance at these particular points to determine the Level Difference (LD) between speech in the
source room and speech at the listening location. The background noise level (LN) is also
measured at the listening location. The SPC value is calculated by adding the overall LD to the LN
to get a value in dB. Both the LN and the LD are measured in 1/3 octave bands and the values at
each 1/3 octave band are linearly averaged to get the overall LD and LN.
The understanding or comprehension of actual overheard speech was tested extensively, as was
the level of speech and noise in typical meetings rooms to craft the new method. The SPL of speech
in a meeting room was found to be 69 dBA or quieter 98% of the time, and as the intelligibility of
speech depends on the signal to noise ratio between the signal (speech) and noise (background
noise), the speech level in the source room is of fundamental importance to the result. To ensure
good speech privacy, the partition between the protected space and adjacent spaces must
effectively attenuate the speech sounds to an unintelligible level. To help with this, the background
noise level outside of the protected space should be loud. Thus any speech that leaks through the
partition will be masked by the loud background noise level. Further, the background noise level
inside the room should be quiet and the reverberation time short. This will help minimize the volume
of the original speech inside the room as talkers will not try to overcome the noise in the room by
raising their voices, which will help minimize the speech leaking through the partition. The following
chart displays some classifications and descriptions of speech privacy levels.
The interpretation of the SPC value obtained is done in terms of how many words or sentences would
be likely to be understood over a given period of time. This measure is quite un-familiar to people in the
building industry as it deals with human perception at the threshold level rather than with the
performance of a building element such as the strength of a steel beam, which is obviously not in the
domain of human perception. The incorporation of this perception aspect into a measure intended for
use in real buildings needs a new way of thinking. One must consider the human aspect and cognitive
abilities to properly interpret the numbers. The difficulty lies in knowing what it means to be able to
understand one sentence an hour or one per day in terms of what this means for speech privacy. How
much is good enough? That depends on the sensitivity of the information, and on who is doing the
eavesdropping. What might be sufficient for one client might be totally insufficient for another. Time will
tell what values of SPC are required for various uses, just as it took decades for the industry to come to
terms with how much STC was required and how to get it. In fact the STC required for even residential
construction is always in consideration for revision. It started at STC 45, and then went to STC 50 in
1985 in Canada, and it might even go to STC 55 in the future.
One also needs to consider that just as it is true for STC, a design or target value for SPC is not at all
the same as the achieved value determined by measurement in the field of real construction. If you
need high SPC values, you had better hire an acoustical consultant with experience in such design.
The last thing you could expect is to get high SPC by simply selecting a partition detail from a catalog
with a high STC rating. All aspects of the design need to be considered to obtain high speech privacy.