New directions for diagnostic and rehabilitation tools in audiology

Authors: Deniz Başkent1

1University Medical Center Groningen

Background: In our recent research, we introduce new directions to make diagnostic and training tools in audiology more engaging and easier to use. I will present two such projects, investigating 1) the use of music for auditory training, and 2) use of new technologies, such as humanoid robots or automatic speech recognition (ASR), for audiological testing.

Methods: In Project 1, we implemented a novel music training program based on learning improvised piano playing (guided audiomotor exploration, GAME). We have evaluated the training effects with adult cochlear-implant users on perception of voice, vocal and musical emotions, and speech, and collected subjective reports. In Project 2, we investigated modified audiological tests with normal-hearing adults, one using a NAO robot as an alternative test interface, and another combining the digits-in-noise (DIN) test with an open-source pre-trained ASR toolkit, Kaldi-NL. We have evaluated test scores, running times, and subjective user experience.

Results: Preliminary results from Project 1 showed the training method to be effective in teaching how to play piano, and also to provide a benefit on perception of speech in noise. Project 2 showed that, despite the potentially small speakers of the robot and no specific training of the ASR, both seem to be capable of producing similar test results to traditional computer-based versions. The processing times, however, could be longer. In both projects, participants reported positive experiences.

Conclusions: With increasing scientific knowledge and improving technologies it is possible to design enjoyable and reliable tests and training tools for hearing device users. Positive experiences may motivate, for example, to continue with training. Some new technological tools seem to already be able to produce reliable results, but need further fine tuning for a range of populations, and reducing processing and running times, to be fully integrated into clinical tests.