top of page

ARCHAEOACOUSTICS: An Interview with Umashankar Manthravadi

This was an interview conducted with Mr. Umashankar Mantravadi by Yashas Shetty. The interview was conducted in January 2021 and explores “acoustic archaeology” and its tools. We're still working on the accuracy of the transcript but this is what we have so far.

Mr. Umashankar Mantravadi recording the Oggu performance in Anupu

Recording the Oggu Performance in Anupu.

Picture courtesy: Mr. Umashankar Mantravadi

Y: What is acoustic archaeology?

U: It’s fairly new, about 25-30 years old. It is acoustics and archaeology coming together.It is related to for example, Architectural acoustics, which is the acoustics of built spaces.

Acoustic archaeology is the study of acoustic properties of ancient structures. It draws much of the technology from Architectural Acoustics. Lots of people have been looking into really old ones, structures 20,000- 50,000 years old, caves from the Stone Age. Most importantly, there is no consistent system underlying this study.

Y: If we consider archaeology, archaeologists have certain tools. What is the relationship between the tools of an archaeologist and that of an acoustic archeologist?

U: They share the tools for measurement, the physical measurement. They measure the dimensions of the spaces. Although archeologists are not very precise with their measurements. The drawings are very elaborate but when you look at the actual measurements, they are not always accurate. They take photographs along with a scale to get an idea of what the sizes might be.

After that it gets diverted because with acoustics, what they are looking for is tiny and very basic surface properties of materials which are much more important than the surface look of things.

Y: So therefore to summarise, there is a shared tool of measurement?

U: so the visual elements are shared. A lot of the the other things like whether it was built in the first century or the seventh century, are not very relevant to me for my measurement.

Y: so therefore dating is not relevant for you?

U: Dating is only to put it in context and I will take the dates the archeologists give me, I will not do my own dating.

Y: what about the style?

U: Style again I will take the style, inscriptions an archeologist and history archives give me because I would need a panel of experts for that.

Y: So that is a separate area of expertise that you have to draw upon?

U: For instance, at one point in the beginning I had suggested, but again nobody took it up and now I am going to do it myself, was to use ultrasonic on inscriptions. There were a lot of inscriptions where the surface is eroding so you can’t read the letters anymore but one of the things ultrasonic can do is that it can give you information.

A lot of people do that but I can’t do it myself as I’m not an expert in the field.

Y: Basically you have said that where the common thread is measurement and the things that you are measuring are for example, the acoustics properties of a space-- the amount of time and reverberation for example?

U: Yes, reverberation is the most common and recognisable of them. There are about eight online measurements we do. Now luckily, we can do a single measurement and then all the eight or nine properties are extracted out of that measurement.

Y: Now is it similar to a fast Fourier transform where you have a wave and you get its properties?

U: Well it is not, It only shares some elements with fast Fourier but actually you are doing it entirely within the time domain. Though we do consider it to create an impulse which is a single pulse which has all the frequencies in a certain ratio. The returning pulse will tell you a lot about the properties of that space.

Y: Now these eight or nine elements, is it possible for you to list what these are?

U: No, I will have to look it up. I can tell you a few but they will not be exact. One of them is reverberation, RT60. The time it takes for a noise to reduce by 60 decibels, 60 decibels is taking into inaudibility. You know if you take a tuning fork and ping it, the ping will take a certain amount of time to fade out. That is the reverberation time and is called RT60. But later on they realised that a single RT60 is not sufficient because in some places the initial falloff is very sharp and then there’s a faint low level sound in the cave that lasts for a very long time and in others it is the other way around. It stays for quite strong for a very long time and then suddenly disappears. So you do three of them, RT20, RT40 and RT60.There are four other measurements like clarity. There is something called the RASTI, rapid analysis of speech transmission index. It is actually a measurement for a way to speak but RASTI measurement tells you about what it will sound like if someone made a speech there. Clarity is measured, gain is measured for how loud the space is. There are other things which are three dimensional. One of the things is reflection. If they come from wide apart sources, you tend to be better at localising sound.You can use the AURORA plugins by Angelo Farina to get a sense of all the measurement parameters.

Y: I will look at this program and maybe ask you to expand on that later. We can talk about a concrete example where you have used these eight properties.

U: I have done it in 5 or 6 places so far and I usually make a chart. Even then it doesn’t because what we hear and how we hear is very different from looking at the chart and saying, “this has a reverberation of 0.067 seconds.” You don’t really see the chart and get a sense of what it is. Whereas you can listen to the recording, that’s when I started using the IR(Impulse Response). The IR itself is a tool for comparison. You get the impulse response from various places and you get it from the source and you can actually see a single sound, process them with each other and hear the difference in the quality of the sound.

Y: In terms of setting up your measurements, what is the process?

U: You set up the loudspeaker. Ideally it is an omnidirectional loudspeaker which is a sound source because with anything that is directional, no two speakers are alike and its difficult to make comparisons. With omnidirectional speakers you can define it, you take a 12 small loudspeakers mounted in good approximation of an omni loudspeaker. All the speakers get the same sound at the same time. That should be put on a stand of about 2 metres in height and you do your measurements at a fixed distance of two or five metres to get enough reflection. If you get too close you’ll only get the direct sound which will sound the same anywhere, some distance is important. You go to a particular spot, the interesting thing that there is a clearly defined space for a stage and for the audience but in some places there is no clear to distinction as to where you should measure so you take measurements from multiple spots. If a place is symmetrical then you take measurements from the left side and don’t have to worry about the right or you end up with too many measurements.

Y: What is the relationship between the physical and acoustic measurements of a space?

U: A standard rectangular space like a shoebox will have direct correlation, in fact you can mathematically calculate what the sound will be like in that space because the models are very pure and reliable. The basis for software for acoustic measurements is that you build a model and you can create the acoustics of the space and compare it with the actual place.

Y: in terms of the mathematical relationship between the shoebox and the acoustic properties, what are the equations?

U: the geometric properties are very easy to measure, for example, you produce a pulse in the room, 1m from the back wall towards the centre of the box at a height of 2m and the cube is at 5m. Now you put a loudspeaker at a particular point and produce a pulse, it’ll travel in every direction, hit the walls and travel back. The time it’ll take will depend on the distance it travels. You can note all the distances and you can derive what the sound pressure will be at that point after 5 milliseconds or 15 milliseconds. If all the pulses came back at the same time, you will get the peak. If some cancel or some increase you will get different readings and signal levels at that point like after a certain time it’ll be louder or softer and overall, each time the sound is reflected it’ll have less energy than before. You can look at two things, the first is the total number of reflections that are happening all over the place and under what time it’ll decay (reverberation time). The other is to look at the patterns it is creating, after 10 milliseconds the sound is mostly high frequencies, the low frequencies. That implies that the walls from which it is bouncing or absorbing high frequencies more than low frequencies. So the properties of the walls are something you can actually define. There are charts that give you hundreds of materials and they give to you in high frequency — how much it’ll absorb and how much it’ll reflect. You can use ready made ones for buildings.


Recent Posts

See All


bottom of page