This type of space is apparently the bread and butter for most acousticians – with most non-acousticians being aware of the necessity of acoustical treatment in such spaces. Other spaces are work for us too, but usually only after some problem persists beyond the ‘experienced’ trials of the construction period. Auditoria also suffer from the fact that most are over-treated for music, but just about right for speech. Here are typical acoustical issues that these spaces face.
Large Reverberation Times and Speech Intelligibility
Reverberation is the fundamental problem that all Paanchvi Pass se Tez people are aware of. You cannot get this wrong and get others right. But while getting this right does eliminate the largest of issues, there are many issues it does not solve. I have worked on projects where a previous ‘consultant’ had set only the reverberation time criteria for spaces, and I had to fight tooth and nail and evidence and auralization and case studies to prove that speech transmissibility depends on other factors too – which they hadn’t bothered to address. We settled for an inbetween solution – where I gave them both sets of data – for speech intelligibility as well as their beloved Reverberation (for them to do whatever they pleased with the latter).
I am amazed at how many times the Royal Albert Hall is brought up in initial discussions for such large spaces. At the time it was commissioned, it had far too much volume for its seating capacity and the echo was so bad that it was known as the only place where a british composer could be sure of hearing his work twice😀 (reference available😛 ). Later a number of clouds were hung from the ceiling to provide better early reflections and that fixed it to a large extent.
Getting good early reflections is vital to achieving good acoustics in large spaces. The traditional RT60 analysis is based on some assumptions that don’t necessarily hold good in all spaces. Also there is a growing crowd of fellows who think a simulation software is all that is needed to get the acoustics right. Yes, these help a lot, but knowing what data is ‘real’ and what part of it is ‘extrapolated‘ is important. I’ve known people who compare two such programs and proclaim that one gives you data down to 63 Hz. May I please add that you might want to take that with a pinch of salt – most diffuse field assumptions are not true at 63 Hz, and so it is tricky to calculate even standard NRC data for that frequency band. Of course, we do measure the response of the LF absorbers we design, but the test conditions are not those typical of measuring NRC data for others. On closer questioning as to how they even managed to procure NRC test data for that frequency band , the program evangelist revealed that they assume that if something provides x NRC at 125, it should give you at least that much at 63 Hz. So this is added to the overall frequency response to make it look ‘more complete’.
This is just one example – it is vital to understand the limitations of the software. For low frequency analysis, some kind of meshing software will work much better than ray tracing. Of course, bear in mind that LF issues are much more pronounced in smaller spaces, so with larger volumes, there’s a decent chance we’ll get away with not doing too much to tame the LF. This doesn’t mean we leave it unattended – I’ve seen metro stations (Delhi) and a 2800-seater auditorium that have low speech intelligibility despite having decent RT60. Fixing the LF in calculated measures will help in such cases.
Also, in many cases, the reverberation values of the spaces are calculated in what is sometimes mathematically more efficient and may not truly model the physical process of sound decay in that scenario. Multiplying a T10 by 6, or a T20 by 3, or a T30 by 2 may not yield exact results. The actual RT60 could be slightly different. Understanding the physical process and mathematical algorithms and the approximations made are vital to making realistic sense of the data thrown out by these programs. There’s also another factor – scattering. Calculating with and without scattering co-efficients can actually make a lot of difference. We have to work within limitations of the amount of scattering data that is tested and available.
This depends on factors such as the HVAC equipment and layout, ambient traffic noise, number of windows and their positioning, and in some cases, general MEP equipment nearabouts. I’ve seen auditoria where an AHU was suspended above the false ceiling of a greenroom – right next to the stage, and there was no door between the two areas. There are other spaces where the green room was only 6 feet away from the compound wall of the venue, adjoining a state highway. Traffic noise freely flowed into the green room and was picked up by the microphones. Our very own Chowdaiah hall here has duct cross-talk issues (sound from music classes in the adjacent rooms comes through to the stage) and also has rain noise infiltration . To Bangalore’s luck, we get rains sharp at 5 in the evening during monsoon – just when all evening events start.
Now apart from letting in noise, there’s a larger issue – most school auditoria will typically be built like a huge wedding choultry – with ample natural light coming in so that they save on electricity. Acoustically, we then have to look at different scenarios – when windows are shut the reverberation is very high, when they are open the reverberation time looks good, etc. An auditorium I worked on had vehicle parking stands closeby, and all those shelters had corrugated sheet metal roof – and this place is on the coastal belt – it rains like mad. The surface area of the shelter is enough to create a fairly loud sound when rain falls. Sometimes the number of windows is so high ( for example a triple height auditorium with three rows of windows) that our acoustical estimates vary rather widely between the two extremes – and once commissioned, we have no control over many windows will be opened or closed at any time.
Also the other condition these windows create is that of perfect planar reflections – leading to flutter and ringing. Most often, due to symmetric design, windows face each other. No, curtains don’t solve all issues, although they might be a better option compared to not doing anything at all! The reflectogram has to be carefully studied to understand these aspects.
These significantly affect the speech transmissibility, and it gets very tricky to calculate even reverberation times. This is especially true of spaces with traditional architecture – churches and mosques. In India, many wedding halls and school auditoria have corridors on the sides – these do not form a part of the audience areas, but are coupled with the existing volume. The effect of the double-slope decay contributed by such spaces can be fairly significant.
Shape of the space(!)
Uh huh…I’ve worked on one auditorium that was so perfectly fan-shaped that we ONLY had to solve reverberation issues there – the shape took care of all other things! The same architects were, however, also coming up with a completely circular space elsewhere. I was lost for words. Yes, we can do things to fix that to a good extent but a lot of treatment is then required to just correct issues that are thrown up only due to the shape of the space. Some issues can never entirely be eliminated in such shapes.
There’s that word again. But really, no space is exclusively optimised for speech – with the current real estate scenario, and we are very realistic about that. The extremes of expectations sometimes vary quite a bit – from hosting a world-class basketball court and badminton court, to holding TEDx talks in the same space. Variable acoustics is the key here. There’s no need to buy expensive panels that can be overturned to increase absorption. It’s far easier and cheaper to construct such panels and most schools have enough manpower to get these mounted or dismantled as required.
To conclude, large spaces typically pose challenges with respect to reverberation, and acoustical simulations of such spaces can provide very good predictions of the end results post construction. However a lot needs to be understood about the assumptions, the test conditions, the limitations and the error margins and material data of such software programs. Empirically or superficially analysing such data could prove to be risky. Acoustics is a young science, but we seem to have a massive set of failed examples to refer to. Those are invaluable in the insight they provide. Getting the acoustics of large spaces right is tricky, especially with the amount of straddling across various intended usages that we have to do, but once done right, it is immensely satisfying to see the design take shape as a space that provides warmth, clarity, brilliance, intimacy and envelopment in the desired proportions.