Hear me out: Let’s improve speech perception for cochlear-implant users

Deaf individuals who gain hearing via a cochlear implant have much greater difficulty than normal-hearing individuals understanding speech in conversations with multiple talkers. Here, the consortium works together with signal processing groups to offer novel algorithms that can improve cochlear implants for better speech perception.

Hearing impairment can have a significant negative influence on quality of life, resulting from loss of speech communication abilities. In children, for example, it can delay development. In older individuals, it can lead to increased isolation and loneliness.

In multi-talker conversations, normal-hearing individuals can hear the voices of each individual talker, and use the voice as a cue to pay more attention to this talker and ignore the other talkers. This perceptual ability greatly enhances their understanding of speech. In this project it is observed that current implants do not allow their recipients to appropriately hear voice differences, which partially explains the difficulties in speech comprehension by implant users. The approach involves optimisation of implant settings, such as, the frequency allocation map and the speech processing strategy. A second approach is to improve signal processing of implants, such as, for better spectral resolution. These will improve the communication skills of hearing-impaired children and adults, increasing the chances for normal development and better integration in society.

The results have shown that the frequency allocation map and spectral resolution can both influence how distinct two voices sound to an implant user. Improvements in signal processing approaches show good potential for benefiting implant users in better perception of speech in multi-talker conversations. Hence, the results provide scientific evidence for new algorithms that can be implemented in cochlear implants by the manufacturers.