Ruhr University. Germany: Almost everyone has a relative who is bound to lose the thread during
family get-togethers and who “tunes out” from the conversation: between
13 and 15 percent of people in Germany are hard of hearing. In a large
group, surrounded by many noise sources, with background noises such as
traffic or echoes, for example on the staircase or in a church, the
affected parties find it particularly difficult to follow the
conversation. Hearing aids can help, but only within a limited scope. By
amplifying sound so that the wearer can hear it, they amplify both
wanted and unwanted sound. The background noise and echoes mentioned
above present a problem.
Headed by Prof Dr Rainer Martin, researchers at the Institute of
Communication Acoustics at the Ruhr-Universität are striving to optimise
hearing aids as well as acoustic technology in general, including for
example the speech quality of telephone conversations. “Our aim is to
eliminate interferences from the signal,” says Rainer Martin. “That
means we must make the device identify and amplify one target source and
tune out everything else.” An additional obstacle is that all this must
happen very quickly, namely in real time, so that the hearing-aid user
does not notice any delays. Accordingly, this very complicated
calculation has to happen within milliseconds, and the engineers
optimise it in an ongoing process.
Their daily tools are algorithms, i.e. calculation procedures based
on the static structure of signals. Several algorithms must come
together for the hearing aid to recognise if the sound that is to be
amplified is speech, music or just background noise. “The device is
supposed to adapt to the hearing situation automatically; we don’t want
the user to have to adjust it manually,” explains Rainer Martin.
His colleagues at the institute are currently involved in several
projects aiming at optimising hearing aids of all kinds (see “Quiz and
survival test”). This includes, for example, a cochlear implant which
enables a person to get an impression of what they are hearing, provided
that the hearing nerve is intact. The optimisation of this technology
is also the objective of the collaborative research centre 823
“Statistical modelling of nonlinear dynamic processes” (speaker
university: TU Dortmund) and the EU Marie Curie Initial Training Network
“ICanHear”, which is coordinated by the institute.