Locations with bad acoustics: The Villa Galli case study

At the beginning of 2018, the Kalisode network produced a live event in the venue of Villa Galli in the Sabina region, Italy.
The client had requested a sound diffusion project for speech and background music, with a demarcation of a dance area and live music inside the same hall for receptions.
The acoustics of the room, with a vault of 5 meters and a reverberation time of more than 3 seconds, posed a major obstacle to the customer’s requests and, above all, a strong discomfort for the conversations between the guests.
In this article, I will analyze step by step how the acoustic predictions and the design work of the electroacoustic system were carried out starting from the venue’s drawings, rendering of the installation and thanks to a survey.
The work was divided into five steps: Acoustic surveys, Calculation model, Room set-up simulation, Acoustic correction interventions and PA system.

 

 

 

  1. Acoustic measurements

 

I started building the acoustic model of the room. We needed it to predict reverberation times
during the event and to design the electroacoustic system. We did not search for absolute values, but we evaluated the sampling trend to understand if this was reliable.

The acoustic measurements were taken with a dodecahedral source, and the room impulse response sampling was obtained with the exponential sinusoidal sweep method.

The response was sampled at different points in the hall in accordance with ISO 3382. The main acoustic parameters were calculated for each sampling and the results were averaged between them.

 

 

 

 

The measurements of the empty room acoustic response could not be taken as a reference. They have anyhow allowed calibrating the calculation model (based on the AFMG EASE software), under the conditions provided for the set-up: 160 people present in the venue.

 

 

Average reverberation time measured in the empty room.

 

 

  1. Calculation model

Thanks to the survey and CAD drawings of the venue made by the other Kalisode members, we built a three-dimensional calculation model. Only the significant details from the acoustic point of view were retained and a coefficient of absorption has been assigned to each model surface.

 

 

Calculation CAD model – empty room.

 

 

The model has been calibrated to validate it with measurements taken in an empty room. The simulations and the measurements appeared to be in good correspondence, so the acoustic simulation model of the room was assessed as reliable.

 

 

Comparison of measured and simulated reverberation time in the empty room.

 

 

  1. Room set-up simulation

 

From the CAD calibrated on an empty room, we added tables and guests to the model. Through one simulation it was possible to know the new acoustic parameters. The measured values were compared with those of optimal reverberation for a room of about 900 cubic meters.

 

 

Simulated reverberation time in room set-up compared to empty room and target values.

 

 

 

We pointed out that the furniture presence and people lead to a reverberation reduction not sufficient to bring the values within the desired range.

 

 

  1. Acoustic correction intervention

 

Based on previous considerations, we tried to reduce the reverberation time in the hall. To do this it would be necessary to intervene on the geometry or the type of materials of the limit areas, but in the case of temporary installations, interventions are severely limited by costs and by the hall architecture. In practice, possible interventions are limited to the possibility of introducing curtains.

An intervention based on 55 square meters of cotton/velvet curtain was proposed and accepted average weight (about 500 g / sqm), spaced 10 – 15 cm from the walls. The curtain position is shown in the following figure.

 

Positioning of curtains in the yellow areas.

 

 

It seems clear that notwithstanding a low-cost acoustic correction, a good result is achieved in reducing reverberation, especially at frequencies from the 600 Hz upwards.

 

 

reverberation times with people and forniture.

 

The proposed correction intervention appears to be the minor intervention that results in avoiding the triggering of the level increase mechanism, also known as “cocktail party effect” (see “Acoustic Comfort”appendix).

 

 

  1. PA system

 

Thanks to the same CAD model, it was possible to evaluate different solutions for the microphone amplification system and for an area where to play and play music, minimizing the impact of this latter setting on the rest of the room.

 

The FOH was composed by a Yamaha QL1 mixer and six channels of d&b D12 and D6 amplifiers.

 

Also in this case reverberation plays a fundamental role, but unlike

the previous case in which the sources were distributed inside the room in the form of people engaged in the conversation,

the sound system has very precise directivity characteristics.

To serve both needs, we designed a system based on 5 distributed loudspeakers mounted in a suspended position on the roof-truss braces with two different operating settings.

 

 

Perspective view – sound system.

 

 

 

 

A simulation of the acoustic level was performed, after the system attenuation and equalization and after setting an average SPL level equal to 75 dB (A). The attenuation for the d&b systems E6 and E12 in the use configuration for speech is shown below in the image below.

 

 

Attenuation of sources to obtain a constant level of 75 dB(A) in the room.

 

 

Map of the total SPL broadband level in dB(A).

 

Distribution of the total broadband SPL level in dB(A).

 

 

The sound level was therefore well distributed and 100% of the values were contained in only 2 dB (A), so that the whole hall was uniformly covered.

The level of  speech intelligibility is now verified in two conditions:

the first with an high level of background noise, imposed at 70 dB(A),

comparable to a situation in which guests are busy talking to each other,

and one in which they are in silent during an announcement.

 

 

 

 

 

 

So with the level set at 75 dB (A) it was possible to guarante

the comprehensibility of simple messages to capture the attention of the guests.

With lower background noise, the STI rises to an average level of 0.60,

that is considered in the literature an optimal value for the transmission of speech

in environments not completely treated acoustically.

 

 

 

 

Here follows the music configuration: an E12 speaker was used (connected to a subwoofer) placed above the dance floor area.
The speaker was simulated with an attenuation of 18 dB, to obtain a level of sound pressure of 95 dB (A).
It was important to evaluate the impact of the system on the rest of the room since we tried to minimize this level as much as possible.

 

 

Distribution of broadband levels in dB(A) – disco time.

 

 

An average attenuation of about 10 dB was observed with respect to the value present in the dance floor area. This value may seem not very high, but it describes well the situation of a medium reverberant environment.

 

 

Conclusions

 

The acoustic wall treatment was a necessary condition to obtain a moderate acoustic comfort, otherwise impossible, that allowed the guests to talk to each other.
During the installation phase, the property of the location required to keep the wall frescoes in view, this meant that a part of the acoustic attenuators was merged on the background area containing no pictorial elements.
The sound system worked correctly both as a distributed speech system and as an entertainment system. Passive acoustic treatment has enabled the electro-acoustic system to function properly and for the comfortable guest presence in the venue.

 


 

 

Acoustic comfort for guests

During the reception, the people sitting at the tables wanted to communicate with each other, normally producing a level between 60 and 66 dB (A) at one-meter distance, this loudness level is necessary to make their messages comprehensible.
Higher values corresponded to a greater vocal effort. Because of the reverberation, the sound generated by the guests has been diffused throughout the environment, at a level directly proportional to the reverberation time. It can be considered that to a signal-to-noise ratio of -6 dB corresponds to an understanding of about 75% of the sentences. Given the reverberation time, the guests would try to turn up the volume of the conversation to overcome the background noise and make themselves understood.

 

 

At the same time, the increase in the volume of conversations would lead to an increase in background noise, with a process of continuous increase of the noise level, up to the limit of the vocal effort.
This process was therefore directly proportional to reverberation, so minimizing environmental reverberation was equivalent to improving the acoustic comfort of the environment.
With the acoustic corrections planned for the installation, the noise level that has an impact on the vocal range (300 Hz to 4 kHz) has been significantly reduced. We proceded from a situation where, for a normal vocal effort of 60 dB (A), we had a signal-to-noise ratio of -7 dB (A) in the untreated environment, at a value of
-5 dB (A) for the treated environment, with a consequent intelligibility definition increase.
The value of -5 dB (A) of the treated environment allowed guests not to raise the vocal effort and consequently to converse quietly.