Article - Issue 40, September 2009
Auralisation - Engineering sound for public places
Rob Harris FREng
Arup’s SoundLab allows architects, engineers, and musicians to hear how different spaces perform acoustically and how architectural form shapes the sound quality. The SoundLab is 20 m2 in area and the ambisonic 3D sound comes from the dozen loudspeakers arranged in a sphere © Arup
Advances in computer technology are now influencing the acoustic design of buildings and open spaces. Rob Harris FREng, director of Arup Acoustics, is a leading proponent of ‘auralisation’ and an eminent acoustic designer. He has used this process in sound laboratories to help create the acoustics of landmark auditoria including the Oslo Opera House and the new Kings Place Recital Hall in London.
The process of listening to a predicted sound field in a particular location through a computerised process is called auralisation. It is the aural equivalent of visualisation.
The process starts with an anechoic sound recording. This is a recording made in a fully sound-absorbing chamber, that ensures there are no sound reflections in the signal, just the direct sound from the source.
The second part of the process is the ‘impulse response’ of the space to be studied. This is the acoustic signature for the specific source and listener relationship. It includes all the reflections that make up the sound field for the listener. The impulse response is obtained by measurement (for an existing space), by computer prediction or via an acoustic scale model.
The sound and the impulse response are then mathematically integrated to create the predicted sound. Creating and listening to auralisations enables us to engineer concert halls that will produce a wonderful sound experience for the audience.
Data is drawn together in one of the Arup SoundLabs in Hong Kong, Glasgow, London, Melbourne or New York. Sound files can be electronically transferred between them, if desired, to pool expertise or resolve problems. With a recent project, one of Arup’s sound engineers was unable to decide the appropriate height of a concert hall reflector in a hall in Tasmania. They emailed the different recordings to the UK and the USA, went to bed and woke up with a set of returned opinions (which were in agreement) for the optimal height of the reflector.
The SoundLab databanks hold the acoustic ‘signature’ of various public spaces and auditoria, ranging from railway stations, airports and office buildings, to playgrounds and the world’s best concert halls. Appropriate sounds and sources of noise can then be recorded or fed in to see how they sound in these venues. Arup clients and other designers are able to compare the sound of the current design with other venues with a known sound profile, or which have an established acoustic reputation.
When designing a concert venue, sound engineers need samples of a variety of musical styles, recorded in an anechoic room that is free from sound reflections. Different types of composition and a range of musical instruments and voices are gathered up. Then sections of the orchestra are selected separately, such as wind or strings for instance, so that they can be ‘positioned’ in a virtual space, section by section, to build up the desired effect for a particular hall or conductor.
Listening to recorded sound in a virtual location offers clients a cost basis on which to take their decisions and clarifies the decision-making process. Before auralisation techniques had been developed, a client facing major expenditure to reduce the noise level of an office block next to a busy road could have been advised that ‘the internal noises should not be more than 5 decibels higher than the NR40 rating at 500 Hz, measured as the L90 percentile.’ Now a SoundLab model can demonstrate how loud traffic noise will be, and the client can be presented with the effect of costed design changes to reduce the noise, without reference to equations.
In a concert hall project, a change to the geometry of a junction between the walls and ceiling would have made the hall appear longer from an acoustic point of view. This couldn’t have been anticipated from the drawings alone.
The hall geometry and the nature of the surfaces are engineered to provide different sound pathways to reach the audience. ‘Early’ sound reflections, which arrive at the listener soon after the direct sound, enhance musical clarity and acoustical intimacy. Later-arriving sounds bring loudness, reverberant sound (acoustic reflections that arrive at a listener later than 80ms after the direct sound) and sound envelopment.
Low frequencies reflect differently from high frequencies. This characteristic can be exploited to enhance or diminish the ‘brightness’ of the acoustic. The audience is unaware of the different paths by which the different frequencies arrive and only aware of the overall effect.
Opera house and music hall design means balancing these different requirements, which may sometimes work against each other. For the Oslo Opera House, which was completed in 2008, the client wanted a full reverberant sound – the sort of sound that they said was ‘good for Wagner and Strauss’. This is common in Northern European opera houses and a mid-frequency reverberation time of at least 1.7 seconds was desired; in southern Europe, the preferred range is usually between 1.1 and 1.6 seconds.
Reverberation is a property of the volume of space. As the volume increases so does the reverberation and vice versa. Reverberation also decreases the more absorption there is. In effect, absorption appears to reduce the volume. Therefore, for the interior surfaces, materials such as timber, glass, steel or concrete are preferable to carpets and curtains – though what actually causes most absorption is the audience.
Reverberation needs to be balanced with clarity and to obtain this mix the sound has to come to the listener’s ears very early. It helps if the room is narrow because there are stronger reflections and the sound takes less time to get from performers to the audience – the balcony fronts in opera houses are often used to make theatres narrower. The client on this occasion had requested a horseshoe shape of three floors that could hold 1,350 people and the city’s planning department had placed a height limit on the roof. The designers and engineers then had to make a decision on how to maximise the limited volume and space available.
The solution that was developed by Arup in conjunction with local acoustic experts Brekke Strand and the Oslo-based architect, Snøhetta, was a dual one. Firstly, the steel load-bearing roof structure was left exposed internally, creating an overall spatial height of 20m. The main structure of the stone clad roof above is thus included in the volume of the hall rather than being hidden behind a false ceiling.
A huge chandelier was built that doubled up as an acoustic reflector and a light source. It weighs 8.5 metric tons, has a diameter of 7 m and is made up of 5,800 hand-cast glass crystals through which shine 800 LED lights. Its particular form scatters and diffuses sound. Using auralisation in the SoundLab, it could be shown confidently that the reflector was a worthwhile investment, and indeed the argument was made to optimise its size, height and geometry.
Secondly, the auditorium walls above the third floor level were stepped back. The room was made wider at the top by creating a technical gallery which cantilevered out over the walls below, giving the hall a T-shaped section. In this way, the reverberation associated with a large volume was obtained, as well as the clarity associated with a narrow room.
Spreading the sound
These two main elements were backed up with a variety of other effective and flexible measures. A ‘wavy’ back wall was constructed out of oak, to avoid the use of flat concrete which can cause havoc with its echoes. The auditorium wall is mostly convex with very tight concave curves. The architects ran with this acoustic solution which spreads sound evenly across the room and created an architectural feature with the waves.
Performance flexibility was added with heavy, motorised acoustic fabric curtains, which are stretched out along the upper walls to absorb sound when microphones and loudspeakers are used. SoundLab auralisations demonstrated the effectiveness of this variable absorption system.
A reflector above the orchestra pit also helps the musicians hear each other and ‘bounces’ the sound of upstage singers over the orchestra into the audience in the main floor seats.
The balcony fronts of 50 mm solid oak are shaped differently around the room so that the side balconies reflect the sound back down towards the audience, while at the back the sounds are deflected in multiple directions to avoid focusing. All the surfaces in the opera house are made of relatively dense materials to control sound absorption. Another advanced analytical technique – Boundary Element Modelling – was used to optimise the thickness of the boundary surfaces. And background sound from the ventilation system is controlled so that the level doesn’t go above 17 decibels, which is quiet enough to hear your own heart beat.
First night nerves
Waiting to hear the first few musical bars to be played in a new auditorium is a tense moment. At Oslo I recalled the site architect’s comment a week or so earlier, as we walked through the 1,000 room building, with hundreds of people working to complete it and millions spent on making the building look spectacular: “Rob, if the acoustics are poor, all this is a waste of time.” In fact, the acoustics have received unanimous praise from singers, conductors, musicians and critics.
Helping sound artists to match their art to the buildings and spaces in which their works are presented has always been an exciting part of my work. Now I am able to use auralisation techniques to hone the results, allowing the sound to be refined with a computer keyboard, not just carpenter’s tools.
One of the most exciting uses of the SoundLab for me has been to recreate the original Gewandhaus concert hall in Leipzig. This was destroyed by bombing in 1944, having been considered one of the great concert halls of its day. The hall’s acoustic properties have been recreated and these can now be used as a benchmark to design new halls for the future.
BOX: SoundLab at work – Florence High-Speed Railway station
Florence station presented a typical requirement for a terminal in that it was necessary to ensure that passengers could hear all relevant announcements in the usually loud environment of a station. The brief was to not only keep passengers informed, but also to direct them in case of emergencies.
Achieving intelligibility was not just a question of loudness, which is easily controlled, but also of obtaining the required early-to-late sound ratio, described elsewhere in the article. This is determined not only by volume, geometry, finishes, occupancy, but also by feedback and time-varying background noise from passengers and trains.
To understand and evaluate the effect of trains arriving or departing, Arup engineers first recorded high speed train arrivals and departures at an existing location. As far as possible this was the unadulterated noise of the train. This was stored and added to similarly pre-recorded station noise – from passengers, services and other background sounds.
The proposed geometry of the station and the acoustic characteristics of the surface materials were input to the computer model. These surfaces were modified in the computer model runs to reduce or absorb the train and background noise and the unwanted, intelligibility-reducing late sound reflections and reverberation.
The next stage of the process, which is typical of many design exercises, was to place simulated loudspeakers at various points in the computerised model. These can be moved or realigned and various announcements can then be made during the simulated sound of trains arriving or departing or standing stationary.
Adjustments are made to the wall and ceiling surfaces and possibly some minor changes made to the interior geometry. Still using the model, announcements are listened to at various points within the station and loudspeakers moved or programmed or both so as to provide the engineer – and the client – with a realistic understanding of their clarity and effectiveness.
Finally, the sound was added to ‘walk-through’ visualisations of the station.
BIOGRAPHY – Rob Harris FREng
Rob Harris began his career as a theatre sound engineer, mixing the sound for major musicals in the West End and Canada. He specialises in the design and refurbishment of opera houses. His credits include the Britten Opera Theatre at the Royal College of Music, Glyndebourne Opera House, the refurbishment of the Royal Opera House in London, the Wales Millennium Centre, the Operaen Copenhagen and Oslo Opera House.
Rob Harris’s most recent projects are the Kings Place chamber music hall in London, for which he was the Arup director for both theatre systems engineering and acoustics and the refurbishment of the concert hall of the Royal College of Music, for which he led the theatre systems engineering team. He was elected a Fellow of The Royal Academy of Engineering in July 2009.
The author would like to thank Fabian Acker for his help in drafting this article.