The cochlear implant is the most successful man-made interface to the human neural system to date. It allows to restore a sense of hearing in people suffering from profound hearing loss by direct stimulation of the auditory nerves. It is a small electronic device surgically implanted in the inner ear that can restore the hearing sensation of profoundly hearing impaired patients by electrically stimulating the auditory nerve. Although speech intelligibility in quiet environments and even telephoning is no longer a major problem today, speech intelligibility in noisy environments is still a major challenge. CI wearers need significantly higher signal-to-noise ratios (SNRs) to achieve the same speech intelligibility as normal hearing people. For this reason, binaural speech intelligibility algorithms have recently emerged that aim to increase the SNR in noisy environments by processing the signals from both the right and left ear on the other side together.
However, the transmission of the audio signals from one ear to the other ear has to be performed with low latency, low bit rate and very high audio quality, so that the binaural algorithms can lead to an improvement of speech intelligibility in noisy environments without unnecessarily reducing the battery life of the devices. This project therefore aims at the development of a novel data reduction method (codec) that allows such signal transmission and thus binaural signal processing for CIs. The codec is based on an efficient transmission of electrical excitation patterns and should be specially adapted to the above-mentioned conditions in order to achieve the highest possible gains in speech intelligibility and speech quality. It will be combined with advanced methods of binaural signal processing and evaluated with respect to its influence on speech intelligibility, audio quality and localization by means of various subjective tests. The intended project is a joint project of the Auditory Prosthetic Group of the German Hearing Center (part of the Hannover Medical School) (MHH) and the Institute for Information Processing (TNT) of the Leibniz University Hannover. During the course of the project, the project partners will be able to make significant progress within their research areas, namely data reduction for audio applications (TNT) and binaural signal processing for CIs (MHH). The collaboration also offers a great opportunity for an exchange of knowledge between an engineering institute and a medical school. By combining engineering methods with expertise in the medical field, the success of the project can thus be ensured.