< Program

Translational Research I

Voice and Speech Perception in Children with Hearing Devices
Deniz Başkent, PhD
Professor, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands

Objectives: While hearing aids and cochlear implants aim to provide best speech perception, there are other important aspects of communication that the hearing devices may also help. Examples are perception of a speaker’s voice or vocal emotions, or segregating target speech from background speakers, enhancing speech-in-noise perception. These perceptual mechanisms are essential for overall development of children. Yet, studies with children with hearing devices can be complicated, due to the interactions of developmental aspects with hearing-related factors. In our research, we aim to address this complexity by conducting studies on voice perception systematically, using speech synthesis techniques in a variety of voice-related tests, testing children with differing hearing device configurations, and taking into account developmental effects by using baseline normal-hearing developmental trajectories.

Design: Our study participants are school-age children (5-6 years onwards) with hearing aids or cochlear implants. Our control participants are school-age children and adults with normal hearing. For producing stimuli with different voices, we use STRAIGHT speech synthesis to vary two voice cues, voice pitch (F0; related to the glottal pulse rate) and vocal tract length (VTL; related to the resonance cavity, and hence speech formants). Using synthesized speech with varying voice cues, we determine the voice cue sensitivity (the just noticeable difference), voice gender identification (voice perceived as male/female), vocal emotion (happy, sad, angry), and speech-on-speech perception (using coordinate-response measure). We use game-like interfaces, or at times even a humanoid robot, to collect reliable data, even with youngest children tested. 

Results: Results with children and adults with normal hearing indicate that, in general, voice-related perceptual mechanisms require many years to develop. However, the age that the scores become adult-like varies across the tests. Even in normal hearing, within the same age group there is a large variation in scores, indicating individual developmental trajectories. Children with hearing aids and cochlear implants also show a large variation per age, with some scores overlapping with children with normal hearing.  For group data, and for some tests, there seems also a developmental trajectory for children with hearing devices. For the few children with hearing devices whose scores fall outside the age-typical range, there is no clear hearing-related or demographic factor that can explain this difference. The exception is the low-frequency hearing thresholds, which show a significant effect in some tests. In general, vocal emotion perception seems to be a great challenge for children with hearing devices, requiring long developmental periods even in normal hearing. On the other hand, the prelingually deafened and implanted children with cochlear implants seem to produce better scores in general than the postlingually deafened and implanted adults.

Conclusions: The overall impression from our many studies combined is that, when age-adjusted, we can see the benefits of hearing devices for children with hearing loss for many aspects of voice perception. Many children perform very well with their devices, and the few who have deviating scores, or the tests that seem especially challenging for this group, need further attention to identify what factors can be improved in hearing devices or hearing rehabilitation. 


Deniz Başkent has had the great chance to work on cochlear implants as a PhD student and hearing aids as a research scientist. This combined expertise on hearing devices led to many interesting projects over the years, not only on speech perception, but also on other aspects related to hearing, such as top-down cognitive restoration of speech, listening effort, voice and music perception, and developmental and aging-related factors. At the Speech Perception Lab (dB SPL), Deniz and colleagues produce both scientific knowledge and tools, such as on vocal emotion perception, and also applications for rehabilitation, such as music training or use of NAO humanoid robots. Observing and being in awe of how a human brain develops in great speed during childhood and how hearing devices can give support in this has been the inspiration for the studies Deniz will be presenting during her lecture.