Dr Josef Schlittenlacher
EPSRC grant: "Machine Learning for Hearing Aids: Intelligent Processing and Fitting" with Richard Turner and Brian Moore.
In the last three years we have developed hearing tests that use Bayesian active learning in order to make the tests both faster and more accurate. Using them in clinical practice, an audiologist can not only test an audiogram, but for example also outer hair cell loss or dead regions, all within reasonable time.
Further research includes deep learning for the signal processing in hearing aids.
We have also developed hearing models for normal and impaired hearing. Our loudness model that is also applicable to binaural time-varying sounds is currently being standardized as ISO 532-3.
Associated News Items