RTJ_vol6_EN
21/24

Founded to promote advances in knowledge and research and development, IHCON is an international conference that provides researchers with an opportunity to exchange information on the latest research ndings and technolog-ical advances in hearing compensation assisted by hearing aids and cochlear implants. Solicited and contributed papers address current topics related to hearing aid research.In the past, the conference had been held once every two years; with the COVID pandemic, this year's conference was the rst held since 2018, with 251 participants register-ing from 19 countries (as of August 6, 2022). Rion gave four presentations at the poster session focusing on signal pro-cessing for hearing aids.Researchers, audiologists, and engineers assembled from all over the world, including authorities in the eld of hearing from universities in Europe and the United States, including the University of Cambridge, Technical University of Denmark, and The Carl von Ossietzky University of Oldenburg. This conference provides an invaluable environment for international and interdisci-plinary discussions with people from various elds, which is one of its major draws. Meals served in the same room as the presentations helped promote interactions.Participation in the conference gave us the opportunity to discuss the research presented by Rion with many researchers and engineers. In three of the four presenta-tions, prototypes were brought to allow participants to experience the processing sounds. Many in the audience stopped by to have a look. Many people stopped to have an in-depth discussion with us.We would like to continue to participate this conference regularly and gather information on cutting-edge hearing aid industry trends. It’s a great opportunity to explore future directions in research and development Rion should pursue.Binaural Adaptive Feedback Cancellation Based on Prediction Error Method and Its Evaluations for Hearing AidsPresenters: Yuno and HirumaWe presented an algorithm for addressing degraded conver-gence performance during acoustic feedback (howling) while maintaining robustness to the entrainment of the conventional PEM-AFC.The proposed method has two functions: howling detection based on interaural level dierences (ILD) in binaural hearing aids and correlation control for adaptive lattice lter with PEM-AFC.Hybrid speech enhancement for hearing aids with deep neural networks and MMSE-LSAPresenter: SugaharaWe built a compact DNN speech enhancement system that compensates for disadvantages of traditional NR while be-ing small enough to be implemented into a hearing aid. We improved noise suppression performance by mixing the two gains obtained by the traditional NR and the DNN-based NR. We evaluated these methods by subjective experiments in terms of an amount of noise suppression and naturalness of speech.Binaural MVDR-IC Beamformer for Real Time Hearable DevicePresenters: Fujisaka, Hiruma, and Yoshitaka Murayama (CEO of Cear, Inc.)The MVDR-IC implemented in the Qualcomm® QCC5144 chang-es the spatial impression of interference (direction of arrival) by tuning the interaural coherence. We made both subjective and objective evaluations of this tuning.Presentation on the balance between these spatial impres-sions and the perceived clarity of the target speech.Report on the International Hearing Aid Research Conference(IHCON 2022)In August 2022, six members of Rion’s R&D Department participated in IHCON 2022 to present their research results and to learn more about global trends in hearing aids.IHCON (International Hearing Aid Research Conference)August 10-14, 2022Granlibakken Conference Center, Tahoe City, California, USAhttps://ihcon.org/Rion’s presentations (poster session)A scene from an IHCON 2022 meeting roomBy Keisuke WatanukiKeisuke Watanuki, Yoh-ichi Fujisaka, Nobuhiko Hiruma, Yuuki Yuno, Reiko Sugahara, and Ryo Sato(R&D Department Hearing Care, Sound and Vibration Measurement Technology Development Group)ParticipantsPortable low-latency blind source separation systemon Raspberry PiPresenter: SatoIn this study, we presented a prototype of a low-latency blind source separation system that can separate four audio signals from a mixture. The system is based on the AuxIVA, which is suitable for real-time applications due to its low computational complexity. To achieve a high portability, we implemented the system on a Raspberry Pi.19

元のページ  ../index.html#21

このブックを見る