Have you ever seen an engineer conducting an imaginary orchestra? We think of scientists working in a lab, but Professor Tapio Lokki, of the Aalto University School of Science in Finland, has spent the last few years visiting concert halls and making meticulous measurements of their characteristics. ‘Karaoke’ is Japanese for ‘empty orchestra' — and in some ways that is what he has created to help his studies. The research could lead to improved building designs and a form of audio ‘Augmented Reality' (AR).
'There has been research on concert halls ever since the most famous ones were built, more than 100 years ago,' he says, 'but it is still a mystery as to why some are better than others. And when a new concert hall is built, it is still a mystery as to how it will sound.'
What makes a good concert hall work? This is the question Prof. Lokki is trying to answer. His research may even lead to a new form of multimedia Augmented Reality, as well as better designs for auditoriums. 'We need more in-depth knowledge, using modelling, psychology, measurement, music aesthetics and acoustics,' he explains, 'to measure and simulate the behaviour of sound waves produced by 100 musicians in a complex physical environment, and the effects for different audience members sitting in different locations.'
'Testing, testing, 1 - 2 - 3'
But acoustic preferences - and even sound quality - are very subjective, so how can we measure such things scientifically? 'We needed both subjective and objective measures,' says Prof. Lokki. So he decided to borrow some ideas from other fields that need to quantify subjective opinions - the food and wine industries. 'We asked listeners to define their own terms to describe the sound quality of recordings from different concert halls - "bass", "clarity", etc. - and give them ratings. This leads to sensory profiles, and preference orders, for each hall.'
However, his team also needed to provide a standard by which to measure these subjective opinions - how to make sure everyone is reporting back on exactly the same sound qualities? This is where the 'empty orchestra' comes in.
'We built a "symphony orchestra simulator" using 34 loudspeakers,' explains Prof. Lokki. Each speaker is placed at an identical location on each concert stage, and plays a studio recording of an individual player and instrument. 'They always play the same piece of music, played by the same musicians, and then we record the overall sound from identical seat positions in each hall - so the only variable is the architecture.' Then, the researchers invited 20 listeners for each study and played them the piece, while jumping from seat to seat and from hall to hall by switching between recordings, 'so we can really compare the halls', he continues.
The team have profiled mainly Finnish halls to date - and have started on auditoriums across Europe. And they are now working on mathematical models to complement this qualitative research.
From simulating orchestras to simulating acoustics
This is multi-disciplinary research, which needs a sizeable team for good results. ERC funding of the PHDVIRTA project has enabled Prof. Lokki to hire experts in various fields: at first, four PhD students, now expanded to include three postdocs.
'Sound is not like light - it exists at wavelengths between 17 m and 1.7 cm - so echo delays, refraction due to corners, and wall vibrations are all factors - and full computer simulation of each hall is still a long way off,' explains Prof. Lokki. 'But thanks to our measurements, we can reproduce the specific effects that reduce bass, for example, and our 3D simulations can show that, say, staircases in specific locations will act as filters and affect speech intelligibility.'
The researchers can produce visualisations of sound energy - tracking reflections, their directions, and identifying the relevant surfaces - and superimpose these on plans and drawings. This could provide valuable advice on the construction of new concert halls, auditoriums and even libraries or shopping centres.
'As another application, we are working on Augmented Reality for sound with the Nokia Research Centre,' he continues. 'Visual AR uses smartphones or Google glasses, but we can use a microphone to turn headphones "transparent" - the opposite of noise cancellation - and apply this to, say, a three-way phone conversation that fades out as you approach the people you are speaking to. Or you could use it to enhance your acoustic environment. In addition, we have provided hundreds of downloads of our music files from the orchestra simulator, and these are now being used to build on our research all over the world,' he concludes.
The project is coordinated by the Department of Media Technology, Aalto University School of Science, in Finland.