For Marton Marschall, a postdoctoral researcher at the Technical University of Denmark (DTU) in Copenhagen, the department of Electrical Engineering’s new Audio-visual Immersion Lab (AVIL) is “a great technological playground” where he and fellow engineers from academia and industry can probe the mysteries of human hearing.
A collaboration between the university and three Danish hearing-aid companies, the state-of-the-art facility creates realistic soundscapes in order to investigate spatial hearing in challenging conditions.
Essentially a soundproofed anechoic chamber, the AVIL is a virtual environment that recreates complicated “sound scenes”, such as a rowdy restaurant or a crowded railway station. This is achieved with 64 loudspeakers that are arranged in the shape of a sphere, thereby surrounding a listener. The speakers are not only positioned above and around the listener, but also below the laboratory’s suspended floor.
“The subject sits in the height-adjustable chair so that we can position his or her ears at the very centre of the loudspeaker array, and the subject can have an interface – an iPad, for example – to interact with the experiment,” explains Marschall. “The experiment could be to determine the location of a sound source, or a speech-intelligibility experiment.”
One of many phenomena that will be investigated in the laboratory is what has come to be known as “the cocktail-party effect”, whereby people with good hearing can focus their auditory attention on a particular stimulus while filtering out a range of other stimuli, in much the same way that a partygoer can home in on a single conversation in a noisy room with multiple sound and speech sources. Hearing-impaired individuals experience extreme difficulties in such environments, even when using hearing aids.
“It’s surprising how well people with normal hearing can communicate in complex acoustic environments such as a busy restaurant, despite the fact that the signal-to-noise ratio can be very bad,” says DTU’s Head of Hearing Systems, Professor Torsten Dau. “It is interesting to understand why people with normal hearing can perform so well, even when compared to a machine, or supercomputer.”
“We also know that hearing-impaired people have tremendous problems, typically with speech communication, in such environments, and it’s not just that they lose parts of their sensitivity to sound – this could be compensated for by amplification. It’s more that; it’s complicated. People have problems in segregating different sound sources, and it’s one of the challenges to understand what is behind these problems.”
In designing the laboratory, KEF LS50 speakers were employed for the central 64-point sphere. “We chose the KEFs because we were looking for a compact studio monitor with a concentric design, and that has two advantages from our perspective,” says Marschall. “One is that the loudspeakers appear more like point sources when viewed from the centre, and that’s nice from a theoretical perspective for sound-field reproduction applications. Secondly, when you are in a sphere like this you have loudspeakers all around you, so it is important that the characteristics of the loudspeakers don’t change, which was one of the advantages of the concentric design.”
Ultimately, experiments performed in the laboratory could have profound impact on the quality of countless lives around the world, while also educating young engineers and preparing them for careers in industry. It is anticipated that the AVIL will contribute to improvements to hearing aids, cochlear implants and speech-recognition technology, the diagnosis of hearing-impairment conditions, and even to the design of school classrooms to maximise speech intelligibility when young children are learning.
According to Dau, about 10 per cent of the adult population worldwide currently have hearing difficulties, and it is predicted that perhaps 100 million people will suffer moderate to severe hearing loss by 2025. The professor adds that while experiments have been undertaken to unravel the cocktail-party effect before, the AVIL initiative is the most advanced yet. “Now we can go for more realistic, more complex scenes,” he says, “where we use sound-field synthesis, ambisonics technology and other methods to go to the limits.”