• AVimmerse

Using Scale Modelling to Assess the Prehistoric Acoustics of Stonehenge.

In the fourth of the Heritage and Technology series, we’re joined by Trevor Cox, Professor of Acoustic Engineering at the University of Salford where he carries out research and teaching focussing on architectural acoustics, signal processing and audio perception.


In his talk, Trevor looks at using scale modelling to assess the prehistoric acoustics of Stonehenge. Listen to what he had to say here:


Speaker transcription*:


Yeah, well thank you very much for inviting me to talk today. So I'm going to talk about some of the work we've done on modelling what the acoustics was like in pre hos historic Stonehenge and particularly concentrating on what you can see in front of me, which is a one to 12 scale model of Stonehenge. Before I launch into that, I should kind of say that what's motivating this work is not so much community communicating heritage, but actually working out heritage. So I'm an acoustic scientist. So my interest here is working out what were the acoustic conditions like in Stonehenge and therefore, provide more information about how it might possibly have been used. So why we need to do this work or making model is because you go back to the current Stonehenge, a magnificent site, wonderful base to go to, there's an awful lot of stones are missing or fallen over. And they're from an acoustic point of view, the sound of modern Stonehenge is not like it would have been in the day, let's say 2200 BC, when a lot more stones were present, it would have a distinctly different acoustic to it. So we can't go back. To go down to Wiltshire. Now we have to find another way of recreating the sound of Stonehenge from the past. And in fact, my interest in this came about from work by a colleague Bruno for Xender, and collaborator group till actually measuring this place here. So this is across in West Western Western America, and it's called the Mary Hill Monument. It's actually it's not the greatest picture of it, but it's actually quite a good reproduction of Stonehenge with a lot of the stones present. I'll talk about why it's not perfect. And they actually went and measured that place to find out what the Stonehenge might have sounded like in the past. And then they recreated it. And actually, they went round, a variety of places recreating the sound of Stonehenge around various museums. This is in drama, one of the collaborators on the project using this device, you can't see the full ring here. But you can see the square ring here of live speakers. This process is called wave field synthesis. And it's is the process they used to reproduce the sense of being immersed in the space. So this is like immersive surround sound, you might have met things like Dolby Atmos or something like that, yourself. This is a more souped up to system. One of the big advantages of this system for creating a sense of being immersed in the space. If you can get money people listening to at the same time, they're both my pictures just have one person, it actually you can have quite a lot of people in the same space and still get a really good reproduction. If you're interested in that particularly correction of left the name of the paper up, then you could look it up and have a read of the paper about what they did on it. Now the problem with those big souped up systems and and when you're trying to do that is they're great in museum contexts, but they're not so useful if you want to do it in somewhere where it hasn't got expensive sound systems available. And of course, you could use head up displays nowadays. And we do a lot of work using head up displays for virtual reality. In our case, it's usually for research playing sounds to people where they're looking visuals and getting to see how they respond. But I just mentioned very briefly, there are some really interesting things around and audio reproduction you can consider. And there's this thing called mediate device orchestration, which kind of project I was involved in, which is, if you think about looking around your home, you zoom most of you at home at this time. We have an awful lot of loudspeaker and sound reproduction equipment in our home. We have televisions, mobiles, Bluetooth loudspeakers, Alexa's laptops, all sorts of things. So you can do some really interesting things, actually just making ad hoc arrays of devices. So you go away from trying to do everything exactly right, with a big, expensive, smart speaker array. But you get a sense of immersion from a set of ad hoc devices. And it's quite an effective, surprisingly effective technique. I thought I'd mentioned that because we're interested in reproduction heritage, this is something that doesn't require people to have expensive sound gear to make it work. If you're really interested in that idea, you can look up BBC taster. And look at all the orchestrator. We've got the BBC tape to look at pick apart this is actually got nothing to do with heritage. It's actually a thing for BBC Three, but it's using the same technology. And you can have a play around it's actually listening to a folk band that you can have different instruments on different devices. So you can have the flute on coming out of mobile phone and the bass coming out of your laptop wherever it might be. So if you want to play with that technology, there is something on BBC taster for the next few months you can have a play around with if you want.



So to return back to my particular project, which is about Stonehenge. So I said I've got two colleagues went and measured Mary Hill, this monument, which was built after World War One. What wasn't that wasn't job done? Well, no, because the Maryhill monument was is not actually a completely faithful reproduction of what Stonehenge was like. So on the right hand side, I've got a picture of Mary Hill on the left, I've got a picture of my model that I made. And the basic thing about Mary Hill is the ground is wrong. The stones are a bit too regular the bit to square. So it's not quite acoustically not quite right. So what one reasons he wants to move on from Mary Hill was to try and get a better acoustic measurement, because my interest is actually trying to find out what the acoustics were like, and then look at how we can make it interesting for people to engage with. So what can we do, so we can make a model, but there is another technique, and we could of course, use a modelling technique. And this is actually an example of a computer model of Stonehenge using what's called finite difference time domain. And you can see it's a bit like trips or referral time when you do physics, we can just imagine how the waves are going between and reflecting off the stones, and giving you a sense of how sand moves between the place. Now the problem we have with Stonehenge is it's very inconvenient thing to modelling computer models. So we have big computer models that are available for simulating, say concert halls and use them to sign a concert halls that work terribly well for Stonehenge because of the stones and the occlusion, the focusing and things going on. And we have what I've just shown you here, which is a wave based model, which is really good and accurate. But you can't get to a very high frequency because it just takes far too long to run. So computer modelling is a really interesting area. And you could use that to recreate the sense of Stonehenge. But if you want to do accurately, currently, we don't have the technology. So it's an interesting area of kind of development. Let me just close the door here. Apologies for this. I wasn't anticipating my dad to start playing the concertina in the background, as I was talking, anyway, to move on. So that led us to look at Matthew making a model of Stonehenge itself. And this is a reproduction of what it was like in 2012 BC. Following the best archaeological evidence, we know at the moment, now I'm not an archaeologist, I'm following whatever was published in papers for the best reproduction, you can give the sense of the size of this thing here. It's a special chamber called a semi anechoic chamber. And you may think this is I mean, this was new to prehistoric modelling. No one had done this before. Before we we did this for Stonehenge, but it's not a new technique to to acoustics. So it's funny to be talking on this sort of kind of talk about modern and new techniques to have a method that goes back actually to 1930s. So who in fact is a pitch for the 1970s, so it might be late 60s, this is the Barbican Hall. And this is Uncle Mike Baron who's quite well known in console acoustics are now retired in a one to eight scale model of the barking concert hall. So this is a technique that's been used quite a lot of time. But it's falling gradually more out of favour as computer models are improving. But we were the first to apply this technique to a prehistoric monument. And as I was saying, the prehistoric monuments Plexi and the scale of Stonehenge isn't very, it's quite difficult to compute a model so what's what do we have to do so we have this Stonehenge on the left docks young got complete one that says a picture and we're gonna go to one to 12 scale. Now we're using we choose one to Tosca is basically we have to fit it within a specialist chamber, this is called the semi anechoic chamber, all those foam wedges absorbs sound. So when sound leaves or stones, it doesn't get reflected back in by the room, which is what it's like a wheelchair around the Stonehenge, the stone part Stonehenge, there isn't much else. So that's what that's simulating. So we ended up with one to 12 scale. And that creates a series of problems actually modelling when you do that. So in room design, we typically work at this frequency range. And if you're not used to frequencies 125 hertz, I don't know what's that going to be some like a trombone playing 4000 hertz piccolo playing, it's that kind of frequency ranges, basically the range of the piano. No, not quite as low as that. But when we want to model it, because we've gotten to a 12 for the size, we have to go to 12 times the frequency. So when we're measuring in this model, we're actually measuring from 1500 hertz, which is kind of middle of speech range up to 48 kilohertz. Now, project 1000 hertz, I can't hear that I sweat bats and dogs are interested. So we actually tested 12 types of frequencies. And that's to make the physics the same in both cases. So there's, there's things that are known that how to make this measurement work, but they are tricky things to try and get to work. And so there's some exactitude needed and how you could use it.



I thought before someone mentioned it on the chat, I'd point out to one to 12 is the scale that was used at Spinal Tap. I thought I'd be getting there first with the mention, though they only use three stones. So I think they weren't really trying very hard. So there's a whole series of methodological issues this raises one of which is you've got to get the materials absorbing at those high frequencies. The same as the absorption happens at the lower frequencies. And that's one of the reasons why you look at my model, it's not sat on grass, because the absorption of grass doesn't scale properly across frequencies. So that's one of the complications is mapping materials is an issue we've got to work with. Another one is sought is a microphone, so we're going to ultrasound frequencies. So you know, you go and buy a microphone in the shop, type of microphone stop at 20,000 hertz, because no one is above that. So you only have a specialist microphones and a specialist sources to work at these higher frequencies. So that's, that's kind of, so there's some challenges in doing it. But it is a technique that is known at least. So how do we actually make it it's quite funny talking about all these virtual realities. And I'm sure there's a lot of work involved in in 3d modelling and those spaces where I remember, there was a lot of modelling to make a physical thing. So Historic England did a scan of Stonehenge, a laser scan, and then they recreated what Stonehenge would have been like in various phases of its construction that were, I believe, was done by University of Sheffield, by Mike path, Parker Pearson's great when he was there. So we started our work from this model. And so this is the CAD model they sent it. So actually, after considerable work, actually separating all the stones out to and then creating them individually. There's a lot of stones it took a lot of months to make. So you think about how can we make this Well, I suppose the obvious first technique to think about is 3d printing. So here we have a blue stone to blue stones are the little stones, I say little they're one and a half two metres tall, but a little compared to the rest of them being printed. Obviously, that speeded up quite a lot. The big, the big stones like the big telethons in the middle, which are I guess it kind of like thigh high waist high, but around that date took about 36 hours to two days to print. So it is quite a slow process to try and make. So 3d printing was one technique we used. We also use moulding, because 3d printing would have taken six months and we didn't have the time. So I learned a lot about how you mould things. When I was a kid I moulded the chest set, this was kind of similar, but on the other larger scale. So this is one of the larger telethons, this is a 3d print. This is a silicone mould made from that 3d print. And then you pour into that mould you pour a plaster polemics to actually make the stones and this is a way of casting and making stones a bit quicker. So that was one of the processes we used. So there's a lot of a lot of work involved as 100 157 stones we created by this kind of process, then you've got a problem of how do you set it up in the chamber. So we had these boards marked up, as you can see with all the locations of the stones, and then you spent a long time gradually putting them into the room.



That's a lot that took a lot longer than they take in a slow lapse video like that. Actually, the putting info, just taking photographs is quite simple. But when you do it for real, you have to seal all the gaps up. So you have to get around with plaster seen similar gaps. And that takes many, many hours. And it's a tedious job, I should be doing it again this August. And I can tell you, it's a tedious job. But there you can go. That's the reproduction of Stonehenge one to 12 scale give you probably got a sense of size, but this is about two and a half metres across here. So that's once you've created that, then you've got to make decisions about measurement, I've got to tell you a few of the measurement results, I won't go into the acoustics in great detail. But the last slide, there's a link to a paper that has all the details of the actual measurement results, we can ask me about them. So we did a lot of like phones over requisitions, we did some inside the circle. Lots inside few outside of green squares, our source positions, down the middle, some of the edges, some behind stone, some front stones, lots of different places, because we don't actually know how to Stonehenge reduce. So we can't say this was where people stood, we have to go and measure different places. But that's those are measurement positions. And I don't know if I'm going to tell you one such scientific graph, but have a scientific graph for you. So this is what's called the reverberation time, on vertically, and this is frequency along the bottom axis. So I said this frequency range is roughly a piano range. It's not quite but it's roughly 225 hertz, low end of the piano 4000 hertz, the top end of the piano. And that's typical design range for architectural acoustics and reverberation time, you would have met very reverberant space. So you could go into a big cathedral reverberation time might be five 610 seconds, something like that. That's how long it takes sound to die away to nothing when you stop. I don't know playing a musical instrument. If you go into a concert hall, like something like Bridgewater Hall in Manchester, then the reverberation time is a bit shorter two seconds. And I'm currently talking to as I might have indicated from the bedroom actually my dad's house. And I guess the reverberation time in here is about naught point three seconds it's quite dead in the space So we have three different lines where the black one is the current Stonehenge. The red one is the Maryhill replica. And the blue one is the new model. And to give you a sense, so, you know, reverberation time the middles, but point 6.7 seconds. What's that? Like? It's probably like a cinema a really. So it's a big space, which is fairly dead. And that's kind of the the centre of it is what we can say is reverberation. Like that would have helped music to sound a bit better. And we reverberation mixing sound better from music. So a little bit would have helped a bit more would have been probably even better. And it would have also helped speaking. So that's what we were looking at how to get help speaking how to help with music and things like that. There's can we get into some sort of interesting archaeological debates about what was going on? You know, when, when people were having ceremonies if they were talking, what how, how quiet were the crowd? Was it like going to the pub where it was lots of noise? Or were they all sitting reverently and hushed that makes a difference to have cotton stones or how was speech intelligibility important? If you go back pre reformation, in churches, you were having people talking and Latins speech intelligibility was not important. And you can see our church acoustics, the design has varied depending on whether the liturgy is in English or not, or native language or not. But anyway, basically, what we find is that it is good for speech, there's lots of reflections from the stones, that means you can you can hear very well across the stones, it amplifies a bit for decibels, it's a little bit amplification, that's useful as well. So within the circle, there's certainly some help to make speech communication easier. Now, thinking about what this talk is partly about linking into his heritage experiences, so one of the things we do a lot of is do organisation, and that's reproduce what the sound of things were in the past, or when you alter it, so it's, it's the sound equipment visualisation. And it's the beta engine we do this is basically WhatsApp behind game audio and stuff like that. But we are working on much more accurate when accurate measurements of accurate space was like when we're doing it. So I thought I'd play you a normalisation and tent, zoom to screw me up. And first of all, we're going to hear some speech without the stones, and then you're going to hit the effects and stones added on. And basically, the stones reflects. And so it's a bit like singing in your bedroom and then walk into the bathroom, you hear a difference in the tone of your voice, a bit of amplification, that's what you should kind of hear even down the Zoom line. So let's get Let's risk it working. You're hearing my voice recorded in an anechoic chamber. There are no reflections, all you're hearing is the sound straight from my voice to the microphone.



It seems to frozen, doesn't it? And here I am in the middle of Stonehenge and all realisation of what I would sound like if I went back to 2200 BCE and talk within the monument. Well, I hope you can hear that I'm sorry about the sort of pause in the middle, I think I clicked on the chat, see what happened, I should pause it. So apologies. That wasn't zoom. That was incompetence of your presenter happening that. So that's the sort of thing we can do. We can reproduce the sound from the past. And we do that quite a lot. And someone says in the chat, this is a technique that's used in design worked on mentioned our acoustics is one, it's the sort of thing we do all the time from search work as well. You hearing my voice. Next steps I thought I just mentioned I'm actually in August, if you're gonna go and do another set of measurements, what we're going to do is measure the effect of occupation. So those measurements will always know people within the circle. So we've made little model people. Here's one on the right. This is a little model person, which we're going to say See what happens when we have people within the circle did the acoustic effect disappear or not? For in terms of reproducing it and heritage, we want to try and get by normals that rendition was mono, it was one channel because of measurement limits. So we need to measure in the spatial audio, this kind of thing that you might have done with VR AR with a microphone array or binaural with a dummy head. So we will be doing that this summer. We'll also be measuring to higher frequencies that organisation works right for speech, but for music, it doesn't work very well. So I'm going to match our frequencies. I've got some gears go. Let's go to that. And the other step is to get computer modelling to work and I've got various people working on that as well. So if you're interested, you could look at using scale model to assess the prehistoric acoustics of Stonehenge. It's an open access paper, or I put my my email earlier in the chat box. You're more than welcome to get ahold of me. But that's my 20 minutes done.


Keith Myers

Amazing Professor Trevor Cox, that was brilliant. So you've given us a a an ability to listen to the past So that's fascinating. I think it's really intriguing. Yeah, there's, there's there's a lot of application, you know, be interesting to see that process being applied to a lot of other spaces as well. You know, there's a lot of acoustic acoustics rather built in to a lot of these sacred sites. They they obviously knew what they were doing when they when they built these spaces for particular reasons. So that was fascinating. Thanks so much and