Case Study

House Of Worship

Objective: In this case study of a house of worship in southern California, A/V RoomService was asked to ascertain the poor speech intelligibility issues of the sanctuary and recommend solutions that would not harm the existing acoustical qualities for the choir and the large pipe organ at the rear.


Microphone Location #2 of 7

Sanctuary with new PA

Discription and Complaint of Acoustical Problem: The sanctuary has a 30' high vaulted ceiling and hard surfaces throughout providing long decays. There is a single speaker mounted at the peak of the vaulted ceiling, above the middle of the congregation, for pulpit use only. Listeners located near the talker have difficulty understanding speech due to the direct signal being heard followed by the delayed amplified signal with room reverberation being heard simultaneously. Listeners behind this area have trouble understanding speech due to the long reverberation times of the sanctuary.

Preliminary Review:

Test Locations: Using the current P.A. system, pink noise was generated and a hand held real-time analyzer was used to quickly find one good spot (mic location #3) in the pews for spectral frequency response and one poor spot (mic location #4). After these two locations were identified, five more locations were determined to obtain a good representation of the entire congregational space. Often, especially for reverberation tests, averages are taken from different room length, width and height locations. However, because all receivers will be seated in the pews, only locations within the pews and at seated ear height (43"), were considered as relative.

Test Methodology: There are many acoustical tests that can be performed, and after subjective evaluation, it was clear that the tests that should be conducted were different than those originally planned. TEF was used to perform all tests needed: Time Delay Spectrometry, Energy Time Curves, Frequency Response Curves, Speech Transmission Index, Real Time Analysis, Maximum Length Sequence and Noise Level Analysis. Except for NLA and MLS, averages were taken for each test at all seven microphone locations. Output levels remained constant for each location. Data for 44 different tests results were collected. Note that there was no congregation present during the tests, only children and airplanes outside at times.

The average of all tests results for RT60 is 1.75 seconds and shows a very smooth decay. This is very desirable for non-amplified, non-percussive musical performances, but not for speech. 2 kHz. was chosen as the center frequency because it is the statistical center of the speech intelligibility range.


Reverberation - Average RT60 representation of microphone locations 1-7

Speech Transmission Index: With this test, we evaluate the direct sound relative to the reverberant sound energy. Though much more detailed information is available regarding each center octave band, it can be seen below that from each location the overall STI is rated FAIR. From a %Alcons view, (the measured percentage of Articulation Loss of Consonants by a listener. % Alcons of 0 indicates perfect clarity and intelligibility with no loss of consonant understanding, while 10% and beyond is growing toward bad intelligibility, and 15% typically is the maximum loss acceptable) the average from all locations is 14.18, or a POOR rating.

Average Speech Transmission Index representation of 7 locations

Noise Level Analysis: In this particular NLA test, we are interested in comparing the ratio perameters of frequencies, levels and delays in the space. A five-minute sample is shown below from microphone location #6 representing the noise floor of the environment. The surge in the middle of the first bar is air traffic from a small airplane fly-over. Note also the spikes just before 18:16, just after 18:17 and after 18:18, these are periodic pops from the P.A. The noise at 18:19 is a creek from the building. It was determined that the signal-to-noise ratio of the PA over the typical amibient noises was sufficient, and therefor, noise isolation did not need to be addressed.


Five Minute Sample of Ambient Noise Levels

Estimated PA Coverage: Pictured below is the estimated sound energy coverage with the new PA system in the sanctuary for a 2kHz. centered one third octave band, at congregation ear level. The number, location and height of the new speakers was modeled in order to determine the best coverage pattern. The bottom of the rendering is the front of the sanctuary.


Estimated 2kHz. 1/3 octave SPL Coverage for Congregation

Conclusions: The results concluded that speech intelligibility was a problem throughout the sanctuary. The results also indicated that the existing reverberation times are of good quality for the type of music performed. It is obvious that good acoustics for speech, which are short reverberation times, are not desirable for choir and organ and vice-versa. Instead of compromising the two with a happy medium by shortening the reverberation times with acoustical treatments applied to the sanctuary, a better approach was possible.

Recommendations: The information in their report was good news to the church board in that the findings indicated that our solution to the speech intelligibility problem would cost less than expected, not impact the décor, nor the good existing acoustic qualities for music. We recommended a new P.A. system over interior acoustical treatments. By introducing many small, localized speakers for the congregation, better coverage was obtained. This introduced improved intelligibility for three reasons:


Results: The church board was surprised with our findings and happy that our recommendations were less evasive to the budget, the decore and the music. The results were so good that no further testing was needed, nor further acoustical treatments. Instruments, choir and pipe organ still have their rich reverberative support, yet speech is now uniformly articulate at all pew locations.