Lossless Compression at Zero Delay of the Electrical Stimulation Patterns of Cochlear Implants for Wireless Streaming of Audio Using Artificial Neural Networks

verfasst von

Reemt Hinrichs, Lukas Ehmann, Hendrik Heise, Jorn Ostermann

Abstract

Cochlear implants (CIs) are battery-powered, surgically implanted hearing-aids capable of restoring a sense of hearing in people suffering from moderate to profound hearing loss. To achieve this, audio signals captured by the microphone of the CI are processed by its signal processor and converted into electrical pulses, the stimulation patterns, which then excite certain areas of the cochlear. Nowadays wireless transmission of audio from external devices, like remote microphones and smartphones, is used to improve speech understanding and localization or for the convenience of the CI user. To conserve energy or channel capacity in this wireless transmission, data compression is commonly applied. In this work, zero delay lossless compression of the so called clinical units of the CIs is proposed and a zero delay lossless codec (ZDLLC) based on artificial neural networks is investigated for this purpose. The ZDLLC is compared to the lossless compression algorithms PAQ and PPM as well as the lossy Opus audio codec. On the TIMIT speech corpus and various acoustic scenarios the ZDLLC achieved a mean bitrate of 28.6 kbit/s at zero algorithmic latency compared to 33.6 kbit/s to 35.2 kbit/s for the Opus audio codec at 5 ms to 7.5 ms algorithmic latency. In contrast, at very high latency, PPM achieved a mean bitrate of 37.3 kbit/s and PAQ achieved a mean bitrate of 25.1 kbit/s. It was found that lossless compression of the stimulation patterns could be useful for wireless streaming of audio.

Organisationseinheit(en)

Forschungszentrum L3S
Institut für Informationsverarbeitung

Typ

Aufsatz in Konferenzband

Seiten

159-164

Anzahl der Seiten

6

Publikationsdatum

2022

Publikationsstatus

Veröffentlicht

Peer-reviewed

Ja

ASJC Scopus Sachgebiete

Computernetzwerke und -kommunikation, Signalverarbeitung

Ziele für nachhaltige Entwicklung

SDG 3 – Gute Gesundheit und Wohlergehen

Elektronische Version(en)

https://doi.org/10.1109/ICFSP55781.2022.9924629 (Zugang: Geschlossen)