Ringing the changes
Ringing the changes - With a ‘classic’ reverberation problem and a limited budget, St Mary’s Church Warwick required expertise and ingenuity to solve its problems. Tim Goodyer reports...
St Mary’s Church Warwick is a regular Catholic church in regional Australia, with a healthy congregation enjoying traditional services, and a modest budget for its technical systems. Designed by local architects Dornbusch and Connolly, and constructed from sandstone from nearby Yangan, the church foundation stone was laid by Archbishop Mannix on 29 February 1920. But in 2009, St Mary’s knew that it was in urgent need of a sound system that worked.
While sandstone churches with high ceilings provide a good acoustic environment for a pipe organ recital or small acoustic ensemble, they readily render speech unintelligible. In this respect, St Mary’s is no exception, with a reverberation time in excess of 3.2s and a heritage listing that restricts changes being made to the building or its 35m-high tower. Called in to help, consultant Peter Patrick, from Toowoomba-based Scientific Acoustics, was faced with a challenging set of requirements that included finding a means of obtaining good intelligibility in a space with a long reverberation time using a sound system that didn’t dominate the visual character of the church and could be removed without leaving any trace – a rule in heritage listed buildings. In addition, the system had to be simple to operate, as there is no resident sound engineer.
He began with an Ease model of the building, which was constructed on site with a Laser measure. The model acoustics were tested and compared to the real building’s acoustics and found to be an excellent representation, so a wire frame of the building was built from them.
As the church’s budget constraints ruled out the use of an electronic beam steered array, Mr Patrick turned to Australian manufacturer ATProfessional, and its ALA line array. The ALA range has been created for any application where critical speaker placement and pattern control is required, with up to 32 element systems available. These use 91mm full-range transducers to support a 120Hz to 20kHz frequency response and 150° horizontal dispersion, while vertical dispersion decreases with array length to less than 10°. The ALA-FS range incorporates frequency and amplitude shading techniques, with a proprietary digital signal processing system designed in conjunction with Peter Patrick himself.
An ALA32-FS was assembled by ATProfessional using the ALA fundamental component assembled in a 4m-high array to Mr Patrick’s specifications. ‘It suited the application brilliantly,’ he reports. ‘It stands on the concrete floor and attaches to the top of the columns where holes are drilled well away from the gaze of any observer.’
Two standard ALA 4 (four-element) columns were required to cover the area beside the pulpit and lectern at the front of the church, with an ALA01T (with a soft-dome tweeter to extend the HF response) as a monitor in the sanctuary. In addition to the loudspeakers, a 16-input/16-output DSP system was required to provide the necessary processing power and simple control panel to drive the system. Mr Patrick chose an Electro-Voice NetMax system because it offered a chance to test the new SysTune analyser system with its plug-in feature. The plug-in allows users to pull the DSP screen directly into SysTune and adjust the DSP parameters such as EQ and time alignment, while observing the outcome in SysTune.
The modelling process included a test run after all loudspeakers were carefully aimed and coverage checked. Ease 4.3 includes a powerful simulation package called Aura, with a routine that uses a hybrid ray trace system to evaluate the acoustic outcome at locations selected by the user. The loudspeaker data supplied by the manufacturer controls the dispersion of sound in the same way the real loudspeaker does. The most distant, and potentially worst seat in the nave was placed in the model approximately 20m from the pulpit and 16m from the loudspeaker arrays. The Impulse response derived by Aura was exported from Ease as a .wav file and analysed in the Easera analysis package. The intelligibility calculation from Easera is shown below. It shows that a Speech Transmission Index - Public Address (STIPa) reading of 0.63 to 0.64 can be expected at this location. STIPa is a development of the Rapid Speech Transmission Index (RASTI) used for room acoustics assessment, specifically for sound reinforcement systems.
Once the installation was completed, the system was tested using an NTI TalkBox, which was positioned in the pulpit ‘talking’ to the pulpit microphone, and an NTI AL1 Acoustilyser running STIPa software. TalkBox is equipped with a precision broadband loudspeaker and features human head-like dimensions to mimic human speech, which is replayed from a solid-state player with a CF-Card. ‘It’s worth noting that there are no knobs, buttons or adjustments of any kind on the Acoustilyser or TalkBox that can be used to monkey with the readings,’ Mr Patrick notes. ‘It can’t be tricked by mis-operation and the results verified the design almost perfectly.’
Observers noted during commissioning that the consultant’s voice was loud enough but unintelligible at the pulpit without the PA system, and that it was clear and intelligible when the PA system was turned on. The numbers produced by the NTI TalkBox and Acoustilyser back this subjective assessment, ‘which, given the price tag of the NTI system, is reassuring’, adds Mr Patrick.
The gain before feedback – the highest level at which the system can be used before encountering audio feedback – presented specific problems. Gates in the DSP were ineffective because the 3s-plus reverberation time means the sound pressure from speech is the same volume throughout the entire church. ‘A handclap near the font opens every microphone in the building – no combination of settings of gate sensitivity produced any worthwhile improvement that would still open the gate for a softly spoken talker,’ Mr Patrick reports. ‘Switches on microphones and pressure sensitive mats under the carpet provided the only reliable means to minimise the number of open microphones.’
The feedback issue was solved using a few tricks that included a Bosch feedback suppressor. This incorporates a frequency shifter but cannot change the frequency response, and was used in the path of the speech only microphones in the sanctuary. Minimal foldback was arranged through the DSP from musicians’ microphones to musicians’ loudspeaker, and a pressure sensitive mat was installed under the carpet behind the altar.
While the ALA 32-FS columns steer the direct sound over the altar where a boundary mic is placed, the mat switch only opens the mic when a talker stands on it and delivers normal speech levels – a talker whose mouth is a good metre away from the microphone.
‘The most critical users were happy with the final result,’ Mr Patrick concludes. ‘The visiting priest used the headset and was beautiful and clear no matter where he stood.’
Published in Worship AVL Asia Winter 2010