External websites we've worked on:
Measuring the Acoustics of Stonehenge
When our ancestors stood within Stonehenge, what would they have heard? Scientific measurements have been taken at the original site in the United Kingdom and also at a full size replica site in Washington state, USA in understand the acoustics of this famous stone circle.
The scientific method applied relies on accurate acoustic measurement techniques available not only through technological advance but also within the constraints of measuring an outdoor site, protected as historical heritage, and affected by modern Human activity in the form of noise. In this particular and fortunate case, the existence of a full size replica of the original monument, provides an invaluable opportunity for the reproduction of the original acoustic environment and its measurement.
Acoustic measurements
Outdoor measurements are commonly hindered by high levels of background noise in the form of traffic (ground or airborne) or natural elements such as wind and rain. Another problem with outdoor measurements is the effect caused by temperature gradients which affect sound propagation, particularly over large distances. Air temperature has an effect on the speed of sound and as such this factor needs to be constantly monitored.
Impulse Responses
For both sites, the goal was to obtain impulse responses at specific source-receiver locations. Measured impulse responses contain all the acoustic information from a system comprised of the enclosure, the position of the source and the position of the microphone. From the impulse response it is possible to derive frequency response functions, determine the decay rate of energy and plot energy time curves (ETC) at the measured positions. These acoustic descriptors of the space are used as an attempt to describe the acoustical behaviour of the space in objective and comparable terms. The measurements are presented in the results section.
As no mains electricity was allowed in Stonehenge, measurements of impulse responses were taken by bursting balloons. This quite simple method relies on the impulsive nature of the sound produced when the balloon full of air is burst with a needle. The resulting acoustic interaction between this impulse and the space is then recorded using a digital field recorder and a microphone. Obviously, this method is heavily marred by a very low signal to noise ratio, aggravated by the fact that measurements are taken outdoor. As will be discussed later, only the first few milliseconds of energy decay are available for measurement and a full length T60 reverberation time has to be extrapolated from this.
To record the data an omni-directional microphone has been connected to a battery operated portable digital recorder. Due care was taken to ensure that levels were high enough above background noise but without overloading the input of the recording system. A number of measurements were taken and are shown in the results section.
For Maryhill Stonehenge in the USA, mains power was available through the use of a generator. Hence, we could measure impulse responses using a computer based system. A soundcard has been used to deal with the AD and DA signal conversion. The output of the soundcard was connected to a 1000Watt power amplifier. The signal was passed through a cross –over filter set to a frequency of 120Hz to separate the low frequencies sent to a subwoofer (for the low frequency) and a dodecahedron shaped omni-directional speaker (for the mid and high frequencies). An omni-directional reference microphone was used to capture the response at various positions within the space and connected to the soundcard which provided the necessary polarisation voltage (phantom power).
The impulse responses (IR) were obtained using the swept sine measurement method, with a length of 10 seconds. The swept sine excites the site at every frequency between 20Hz and 20kHz.
Sound measurements at Stonehenge, UK
The original Stonehenge site is located in the Salisbury plains in Wiltshire, UK. The site is located in a shallow valley with no significant geographical features (high mountains, deep valleys, etc) apart from a nearby road. Access to the site on Salisbury plain in the UK was kindly granted by English Heritage. The visit took place at sunrise in May 2009 before the site is open to visitors.
Part of the sarsen circle at stonehenge (photo by Simon Wyatt)
The Stonehenge site is in ruins. All around, many of the stones have toppled over and lay flat on the floor. None of the circles is intact, but there is a substantial perimetral half-circle still standing. The ground surface is turf which is regularly trimmed. No other features exist inside the circle apart from the existing stones. A section of the outer sarsen circle with lintels in shown in Figure 1.
Measurements of the impulse response within the circle have been taken and are now presented. Figure 2 shows the impulse response obtained by popping a balloon in the centre of the space and capturing the acoustic response with a microphone also positioned at the centre of the space. The areas where evident reflections appear in the impulse response are indicated using dashed lines.
Figure 2 – Impulse Response measurement at Stonehenge UK using the balloon burst method. The Balloon is positioned at the centre of the circle and the microphone is also at the centre of the circle.
An energy time curve (ETC) is obtained by squaring the impulse response and converting linear levels into relative dB levels – see Figure 3. The plot allows for identification of strong reflections in the response. The time of arrival and level of reflections can be identified in order to establish any salient features such as a train of reflections at regular intervals which would reveal a flutter echo effect – this has not been observed in the measured data.
Figure 3 – Energy Time Curve extracted from Impulse Response measurement at Stonehenge. Source and receiver are both at the centre of the circle.
In general, the response shows a relatively high density of energy up to around 20ms. There are areas of clustered reflections at 20ms, 40ms and 80ms. Assuming a speed of sound of 344m/s this corresponds to reflection paths of 7m, 14m and 28m which roughly correspond to the radial distances between the centre of the circle to the blue stone circle, the trilithons and the outer sarsen circle respectively. This dimensional analysis is interesting, since the level of arrival of these reflections is at -50dB relative to direct sound making them too quiet for most types sounds. For percussive sounds, with fast transient behaviour, the reflection become more audible. Reverberation appears inexistent from a perceptual point of view.
To determine the wideband reverberation time in the space, an inverse Schroeder integration has been applied to the impulse response data. This is plotted in Figure 4.
Figure 4 - Schroeder integration curve extracted from measured impulse response at Stonehenge UK. Source and receiver are both at the centre of the circle.
The energy decays rapidly, particularly during the early part, as would be expected from an outdoor site. Due to the very short decay of energy in the space, the RT is extrapolated from the -5dB to -25dB decay rate. The presence of part of the circle and scattered stones still provides enough reflected energy to sustain a decay rate in the order of 0.4s.
Acoustic measurements at MaryHill, USA
A replica of Stonehenge exists in Maryhill, Washignton state in the USA. The site is quite remote, with one dwelling and a souvenir shop nearby. The replica was built as a monument to the soldiers of the World Wars and is made out of concrete. The stones, especially the large ones in the outer sarsen circle, have been shaped to appear rough, similar to stones. Figure 5 shows this.
Figure 5 - Inside Maryhill stonehenge, WA, usa. looking left from the altar stone
The Maryhill Stonehenge is sited on the banks of the Columbia River near the border between the states of Washington and Oregon, USA, on the Washington side. Being so close to a large mass of water there are large temperature gradients throughout the day, particularly if the weather is sunny.
We first arrived on site late afternoon on the first day. There was a strong wind that made it impossible to take any reliable measurements or to actually hear any acoustic effects inside the circle. The strong winds are caused by the temperature difference between the warm ground that had been under the intense heat of thesun throughout the day and the surface of the water in the large river. This caused a strong wind rising up the river gorge into land. Next day, at dawn, there was no wind and measurements were possible. We carried out the measurements on two consecutive days, travelling to arrive on site at dawn. Measuring during these early hours not only prevented the strong winds affecting the measurements but also provided a quiet time before any visitors started arriving. The site was open to the public. The signals and equipment used for measurement caused a myriad of interesting reactions from visitors.
During the measurement days, temperatures varied between 20°C at 6.30 a.m. and 30°C at around 12 noon, by which time the strong winds would make measurements impossible. We had to register temperatures throughout the measurement period as this affects the speed of sound and thus any extrapolations from the measured data.
To provide power to the measurement equipment a generator was hired from a local company in Seattle. Since it was a combustion engine the generator was noisy which would influence the measurements if not addressed. As the site was on the top of the river banks, the generator was positioned some 30 metres down the bank to shield the site from its noise. This worked well and the generator noise could hardly be heard inside the stone circle.
Measurements were taken at various positions within the stone circle at Maryhill. Figure 6 shows a diagram of the measurement site.
Figure 6 - Diagram representing the Maryhill Stonehege monument in Washington, USA. The size of the monument matches that of the original site in the UK. The non-shaded stones represent those that are still in their positions in the UK.
The same balloon method employed at Stonehenge, UK, has been employed here for direct comparison and validation of the data obtained at Stonehenge. Figure 7 shows the response with the source and microphone placed at the centre of the circle.
Figure 7 - - Impulse Response measurement at Maryhill Stonehege, USA, using the balloon burst method. The Balloon is positioned at the centre of the circle and the microphone is also at the centre of the circle.
In comparison with Figure 1, there is a clear increase in terms of energy, as expected. The fact that the Maryhill site is composed of full circles construes to a much stronger containment of energy within it. Similarly to the response measured at Stonehenge, clear reflections can be seen at around 20ms, 40ms and 80ms as well as at a few other distinct reflection times. The similarity between this response at that obtained at Stonehenge not only validate the method employed but also increase the confidence that the replica site at Maryhill is a reasonably accurate representation of Stonehenge as it might have been when in its original state.
Figure 8 Energy Time Curve extracted from Impulse Response measurement at Maryhill Stonehege, USA. Source and receiver are both at the centre of the circle
Observation of the ETC obtained from the impulse response, plotted in Figure 8, reveals a similar pattern in terms of reflection arrival times. However, it can be seen that the direct sound and very early reflections, within the first 10ms have overloaded the input measurement system. Extrapolation of levels from this data would be misleading.
As mains power was available during the measurements in the Maryhill monument, the more robust method of obtaining impulse responses from swept sine measurements was employed. Figure 9 shows the impulse response obtained with this method. Source and receiver positions Similar to the cases described above are displayed.
Figure 9 - Impulse Response measurement at Maryhill Stonehenge, USA, using the swept sine method. The Balloon is positioned at the centre of the circle and the microphone is also at the centre of the circle.
The density of reflections is shown in the ETC in Figure 10.
Figure 10 - Energy Time Curve extracted from Impulse Response measurement at Maryhill Stonehege, USA, using the swept sine method. Source and receiver are both at the centre of the circle.
As expected, the ETC results are identical to the ones obtained with the balloon measurement technique. In general, and in addition to the large density of reflections within the 10 to 15ms, there are distinct reflection clusters at 20ms, 40ms and 100ms. It is interesting to note that the energy levels are expectedly higher due to the larger amount of reflective surfaces in the replica site. Reflections at around 20 ms, 40 ms and 100 ms have a level between -30db and -35dB relative to the direct sound. These higher levels suggest that reflected energy may now be much more noticeable. The reflection at around 100ms is particularly likely to be noticed as a very short echo.
Figure 11 - Schroeder integration curve extracted from measured impulse response at Maryhill Stonehege, USA, using the swept sine method. Source and receiver are both at the centre of the circle.
Determination of RT from the current measurement is shown in Figure 11. Interestingly, its value is estimated at about 1.1s in the centre of the space. In comparison with results from Stonehenge UK (Figure 2), this estimation of a longer reverberation time is associated with the larger number of reflective surfaces and the fact that the circles are now complete, effectively retaining the reverberant energy within the circle for a longer period.
Figure 12 shows the Energy Time Curve for a source placed at the centre of the circle and the measurement microphone at the entrance of the circle (on the path between the slaughter stone and the altar; see Figure 6). In this case no distinct clustering of reflection times is evident. In comparison with Figure 10, the focusing effect of standing at the centre of the circle is lost. Reflected sound arrives in succession with no quiet periods between clusters of reflections. Other measurements obtained at different positions within the circle exhibit a similar response in terms of reflection pattern, that is, no focusing effect and clustering of reflections is observed. The particularity of the response at the centre of the circle, when compared to other measurement positions, is evident through the clustered arrival times of reflections. This goes some way into demonstrating that the perception of sound at the central position might be strikingly different to that at other positions in the circle.
In contrast, the levels of reverberation remain generally the same, at about 1.1 seconds, for various measurement positions. Compare Figure 11, Figure 13 and Figure 14, corresponding to centre, edge in front and edge to left hand side measurement positions respectively. For the three measurements the source is always at centre. This suggests an acoustic response similar to a diffuse field within the space. That is, no matter where you are in the circle the rate at which the sound decays is perceived to be generally the same.
Figure 12 - Energy Time Curve for a source placed at the centre of the circle and the measurement microphone at the entrance of the circle. Obtained from measurement at Stonehege, USA, using the swept sine method.
Figure 13 – Schroeder Integration Curve for a source placed at the centre of the circle and the measurement microphone at the entrance of the circle. Obtained from measurement at Stonehege, USA, using the swept sine method.
Figure 14 - - Schroeder Integration Curve for a source placed at the centre of the circle and the measurement microphone at the edge of the circle. Obtained from measurement at Stonehege, USA, using the swept sine method
Discussion
Differences in reflective surfaces and material
One of the first questions that get asked is ‘how much does the difference in material between stones (at Stonehenge) and concrete (at Maryhill) have in the acoustic response?’ In my view, not much. Both materials are quite massive and highly reflective over the frequency range of interest. The stones at Stonehenge are naturally less regular in shape which causes some mid to high frequency diffusion. However, the concrete stones at Maryhill have also been worked to have non flat surfaces. Irregularities are in the order of 10cm deep in both sarsen stones as in the concrete ones. Because they are more regular, diffusion from the concrete ‘stones’ may only exist at higher frequencies when compared with Stonehenge. If this is the case, then Maryhill would be a ‘worst case scenario’, where more focussing and higher levels of reflected energy exist. At the lower frequencies, both structures are acoustically similar.
Through different particular routes however, both surfaces will cause a fair amount of diffusion at mid to high frequencies on the impinging sound waves. The net result within the circle is likely to be the same. In sum, any patterns that were likely to exist at the original Stonehenge site will be revealed at Maryhill with similar objective measures such as reflection times, reverberation and interference patterns or resonances.
Some further investigations are currently being undertaken in order to determine the reflection coefficient of the Stonehenge stones.
Reflection patterns
Both sites show evidence of reflections associated with propagation paths of around 15 and 30 metres when measured with source and receiver at the centre of the circle. Given the physical distance of the inner (blue stones) and outer (sarsen ring) stone circles, there appears to be a clear association of these prevalent reflections as being caused by these two rings.
From psychoacoustic testing, it is known that a single or a cluster of reflections arriving with a delay longer than about 20 milliseconds is typically perceived as an echo. Particularly if no other ‘masking’ reflections exist to ‘fill in the gaps’ between the strong reflections. On the other hand, the relative level of arrival of reflections is also important in their perception. If a structure ‘focuses’ much of the reflected energy into one point, then the level may rise by about 15dB. This is a well known acoustical problem in concert halls with semi-spherical roofs.
From the responses measured at the centre of the circle, the existence of such clustered reflections in the region of 20 to 100ms suggests that they would be audible either as early reflections (arriving within the first 20ms), supporting the direct sound, or as echoes (arriving after 20ms), generating what Humans of the era would perhaps find as an ‘unusual’ acoustic effect. Given the location of the original site, in a wide open plan, it is unlikely that Humans, entering this space would not become aware of its characteristic and unusual effect on sound, particularly if they stood in the centre.
Reverberation
The measured RT in Stonehenge, extrapolated from the T20 curve is around 0.43 s (see Figure 4). It is not surprising that such short decay of energy is measured given the absence of a roof and the decline of over half of the outer stone circle. In its current state, such an outdoor setting does not sustain a reverberation which is worthy of note. The reflection patterns are interesting as they show the same clustering of reflections as was found in the completed replica. In terms of perceptual effects, only the first cluster at about 20ms is likely to be perceived as an effect. Indeed, only with percussive sounds, such as clapping, a faint very short echo could be heard, similar to the effect you get from a building facade. Further extrapolation of results from the site in ruins is therefore of little interest.
The measurements taken at the complete replica site reveal interesting points: The same pattern of reflections can be measured using both the balloon as well as the swept sine measurement methods; There is an expected higher energy density due to the larger number of enclosing surfaces; If measured at the centre, reflected energy is focussed and arrives at the listener with levels between -30dB and -35dB relative to direct sound. This is loud enough to be clearly perceived. In positions other than the centre there are no focusing reflection effects and the perceptual effect is lost although a sense of reverberation still exists; Reverberation Time is of the order of about 1s and fairly constant across the space which may indicate close to diffuse field conditions, at least in the medium frequency range. This level of reverberation is clearly noticeable and would perhaps only be found in natural geological features such as caves.
Sound of Stonehenge in context
The results obtained reveal interesting material for reflection! One cannot ignore the fact that during the era when Stonehenge was being used, there were no similar buildings, at least of that dimension and made of hard reflecting stone (although smaller stone circles were common).
The existence of a 1 second Reverberation Time would certainly be noticeable to any person entering the circle. Interestingly, 1s is a typical, optimal RT for a large lecture hall, ensuring good speech intelligibility. Anecdotally, the space exhibited this feature, i.e. speech was clearly audible regardless of speaker and listener position, undoubtedly due to the large number of reflective surfaces surrounding them but also due to the high degree of scattering provided by the interspacing between the stones, preventing any strong reflections from becoming a nuisance when interacting with the direct sound from the speaker.
This was the case even when positioned in the absolute centre of the circle. There didn’t seem to be a strong effect from the focussed reflections, albeit the difference between central and other positions within the circle is evident in the measurements.
A number of audio samples are available to test this effect. If you are interested email me at b.m.fazenda@salford.ac.uk
Links
Much has been written about Stonehenge, the megalithic monument sited at the Salisbury plain in England, UK. See the following links for some examples:
- http://www.solvingstonehenge.co.uk/
- http://open2.net/timewatch/2008/stonehenge.html
- http://www.open2.net/blogs/historyandthearts/index.php/2008/11/07/timewatch?blog=14
This is a report on a project I have undertaken together with Dr. Rupert Till, from the University of Huddersfield. The aim was to investigate the acoustic sound field at Stonehenge. A good overview of the context, motivation and methodology for this work can be found at http://soundsofstonehenge.wordpress.com/.