Adaptive Sound Field Control

Most modern sound reinforcement systems used for concerts are based on the line array principle, which allows for the control of directivity of the sound radiation of high and mid frequencies. However, the radiation of low frequencies cannot be as easily controlled, as sound waves at these frequencies are less attenuated by air and reflections from boundaries and are damped the least by the structures of residential buildings. Low frequencies are therefore the most critical frequencies in the noise problem of outdoor concerts.

The MONICA Adaptive Sound Field Control solution is designed to control the sound field restricted to low frequencies over large areas with a feasible number of loudspeakers. As such, the solution leverages on the principles of IoT sensor networks provided by the MONICA Crowd and Capacity Monitoring solution as well as the MONICA Cloud server with decision support and feedback communication tools provided by the MONICA Crowd Management and Communication solution. It may further be augmented with the MONICA Collective Awareness Platform for interaction with citizens and neighbours.

The ASFC system interacts with the other MONICA solutions in two ways:

1. The MONICA Crowd and Capacity Monitoring solution provides various collected sensor data (e.g. weather condition and sound pressure) to the ASFC system upon requests. These data are used to update the sound propagation model and estimate the sound propagation in and around the venue.
2. The sound propagation model supplies information on the sound condition in and around the venue to the MONICA Crowd Management and Communication solution in the form of a Sound Heat Map.

The figure below shows a schematic of the information flow in the acoustic closed loop system. Compared to a traditional sound reinforcement chain, a Sound Field Controller is inserted as a processing unit between the mixer and the loudspeaker system. IoT enabled microphones and weather sensors distributed throughout the venue and the control areas continuously measure the sound pressure field created by the ASFC and current weather conditions. These data are made available to the Sound Propagation Module via the IoT platform. The Sound Propagation Module uses the data to estimate the transfer-functions between sound sources and the control areas, which are needed by the Sound Field Controller‘s optimization routine.

Concept of the ASFC (Adaptive Sound Field Control) system

For developers:
 
🛠 The Sound Field Controller
🛠 Results from AFSC tests
🛠 Audio wiring for the AFSC

The MONICA Adaptive Sound Field Control solution can be divided into six parts:

• The Adaptive Sound Field Controller drives the secondary source loudspeaker array by feeding the organizer’s Public Address (PA) system into an acoustic closed loop system.
• A Sound Propagation Module uses environmental data to estimate the transfer-functions between sound sources and the control areas, which are needed by the Adaptive Sound Field Controller’s optimization routine.
• The Sound Heat Map calculates an estimation of how the sound is propagating through a concert venue. A Sound Heat Map can be a coarser prediction for visual inspection or informative assessment.
• Sound Level Meters are used to measure sound levels at strategic locations and send the information to the Sound Heat Map as well as to sound technicians and possible to external stakeholders such as public authorities, the audience, and the neighbourhood.
• A secondary source Loudspeaker Array System is placed behind the audience between the primary sources and the neighbouring region in which the sound from the event should be reduced (dark zone).
• A Quiet Zone is a small localised noise barrier, which creates the highest possible attenuation of noise across the whole listening spectrum so that event staff can speak together in the middle of the concert area.

Adaptive Sound Field Controller: The ASFC system receives audio signals from the concert’s PA console or from behind the PA’s power amplifiers. The latter would be the primary choice because the standard PA system incorporates signal processing units into the power amplification stage. For example, devices such as limiters cause nonlinear components which will decrease the performance of the Adaptive Sound Field Controller. The audio signals, that are needed to feed to an Adaptive Sound Field Controller algorithm, will be the “reference” signal. In all the pilot demonstrations, reference signals were either obtained from the console (mixer) or from a signal “sniffer” device (inserted between power amplifier and speaker channel). Good results were obtained with both methods.

The infrastructure of the Adaptive Sound Field Controller consists of signal gateway/router (signal monitoring and control of main PA system), hub (audio interface), ASFC Core (Digital Signal Processor – DSP), and loudspeaker system (multichannel amplifier and loudspeakers). The ASFC Core calculates the optimal solution based on a sound propagation model estimation. Multichannel signal rendering (filtering) may be processed using a general desktop PC but can be developed as a stand-alone DSP for more efficient computing.

A well performing ASFC system will enable a high sound pressure level (SPL) in the bright zone relative to the sound pressure level in the dark zone. A performance indicator for this reduction in SPL is the insertion loss (IL), i.e. the reduction in decibels provided by the ASFC system within the dark zone compared to the bright zone. With the MONICA Adaptive Sound Field Control solution, the demonstration pilots have obtained up to 10dB reduction in the low frequency band at the various demonstration sites.

Sound propagation model: Sound field control in outdoor spaces requires accurate estimates of the transfer functions between source and receiver. Feedforward approaches are based on direct measurements of the transfer functions in a dense grid of points. However, this makes them intractable for large-scale situations. Hence there is a need for propagation models to characterize the sound field in large areas and provide the ASFC system with such proper transfer functions.

During the first year, the MONICA project investigated the adequateness of Nord2000. The Nord2000 specifies a calculation method for the prediction of the attenuation of sound during propagation outdoors. However, it turned out to have severe limitations when modelling sound propagation at low frequencies and titled to further research and development of alternatives.

During the second year, a new model based on spherical harmonics was successfully tested under controlled conditions in the anechoic chamber for a scaled setup of 2 x 5 m. The acoustic contrast produced when using this model is very similar to the one that can be obtained using a dense grid of microphones.

In the next phase, a novel Source Separation/Contribution technique was introduced in order to improve the estimation capability of the sound propagation model. The aim of this technique is to estimate the amount of noise contribution (in sound pressure level in dB over time, or similar) that originates from the actual concert in the presence of background noise created from other noise sources (such as traffic noise, people talking, etc.). The technique thus answers the question: Is the noise coming from the concert or from other sources, and if this is coming from the venue, what is the contribution from the venue. The Contribution technique was successfully demonstrated at the Sound2019 in Roskilde and may be used for future sound propagation models.

Sound Heat Map: The Sound Heat Map provides an estimate of the SPL at other positions than the one being measured by a Sound Level Meter. The forward sound propagation model calculates an estimation of how the sound is propagating throughout a well-defined area. The forward sound propagation model can be used the calculation of the EU noise indicators Lden and Lnight, in addition to the production of Sound Heat Maps.

The. Sound Heat Map can be used as a stan-alone solution or as part of the ASFC system. The difference is that the ASFC needs very precise predictions on how the sound propagates, while the Sound Heat Map can provide coarser prediction to be used for visual inspection or informative assessment.

IoT Enabled SLM

Sound Level Meters: The transfer-function measurements can be been conducted with standard wired microphones. However, at the concert venues, where the dark zones are far away from the stage and ASFC system, it is recommended to use the MONICA IoT enabled Sound Level Meters described in the MONICA Crowd and Capacity Monitoring solution.

Loudspeaker Array Systems: The PA amplifier output signal is fed into the ASFC system

AFCS secondary source loudspeakers on roof of Tivoli Concert Hall 2018

aas a reference signal. For real-time application, the music signal, or “sniffer output”, has to be convolved in real-time with the Finite Impulse Response (FIR) filters for each control loudspeaker. A sniffer device for converting the loudspeaker amplifier output signal to line levels has been developed. The loudspeaker systems (secondary sources) are extended by the use of additional low-frequency loudspeakers. The secondary source loudspeakers are placed behind the audience between the primary sources and the neighbouring region where the event sound should be reduced (dark zone). It is important, that the secondary source loudspeakers do not negatively impact the sound experience in the bright zone. This restriction must be taken into account in the loudspeaker configuration by using either directive loudspeakers facing away from the bright zone or in the formulation of the loudspeaker signal optimization problem.

Quiet Zone: The principle of the Quiet Zone is to make use of a) the ASFC to cancel out low frequencies and passive elements (sound blocking screens) and b) to block higher frequencies using a passive noise barrier. The aim is to obtain the highest possible attenuation of sound level across the entire listening spectrum without disturbing the sound field from the primary sources. The active part of the Quiet Zone system consists of arrays of loudspeakers that synthesize the cancelling sound-field to cancel out the noise-field in the desired zone. In simulations, different aspects of the design of the Quiet Zone system have been investigated.

Quiet Zone Visualisation

A Quiet Zone, which is quiet in a defined area but introduces noise into the surrounding, lacks meaning. The design is constrained by practicalities, costs and its impact on the surroundings, which means that unwanted acoustic energy could be emitted by the Quiet Zone system itself.

The positioning of the secondary sources is e.g. limited by the dimensions of the speakers – which means they can’t be put very close to each other. On the other hand, having them far away from each other would just be impractical. The number of speakers and microphones should be as low as possible in order to minimize the costs. As a result of the simulations, the following can be recommended:

• A distance of about 1.5 to 2m to the secondary sources is recommended, which is also practical for the physical implementation of the system.
• The optimal distance between the secondary source speakers is about 1-1.2m. A deviation from this optimum does not have a big impact on the attenuation.

The number of secondary sources has a big impact on the costs of a quiet zone system. The number of speakers also depends on the frequency range of interest. For frequencies up to 200 Hz more than three speakers do not have a significant impact. For frequencies up to 500 Hz more than 5 speakers only result in a marginal increase of attenuation.

A passive noise barrier is added in front of the secondary source loudspeaker to block higher frequencies. However, the existence of the passive element might introduce problems into the active part.