Bone-conduction (BC) headphones enable listeners to hear sounds through BC while leaving the ear canal (EC) open to enable surrounding air-conducted (AC) sound to pass through at the same time. However, the intelligibility of presented speech using BC headphones is degraded by BC transmission, especially in noisy environments. This paper proposes a method for improving the word intelligibility of presented BC speech under noisy conditions. The method consists of two types of emphasis: higher-frequency emphasis and consonant emphasis. In the higher-frequency emphasis, frequency components attenuated due to BC transmission were compensated by the inverse-filtering of the transfer function obtained from the regio-temporalis (RT) vibration or the EC radiated sound. In the consonant emphasis, consonant sections with 20-ms short-formant trajectories of subsequent vowels in speech signals were locally amplified by a constant gain. The results of word intelligibility tests showed that both types of emphasis had significant improvements in comparison with no-emphasis. Moreover, we found that the proposed method had the best improvements under all conditions.
Cite as: Toya, T., Zhu, W., Kobayashi, M., Nakamura, K., Unoki, M. (2022) Method for improving the word intelligibility of presented speech using bone-conduction headphones. Proc. Interspeech 2022, 759-763, doi: 10.21437/Interspeech.2022-10463
@inproceedings{toya22_interspeech,
author={Teruki Toya and Wenyu Zhu and Maori Kobayashi and Kenichi Nakamura and Masashi Unoki},
title={{Method for improving the word intelligibility of presented speech using bone-conduction headphones}},
year=2022,
booktitle={Proc. Interspeech 2022},
pages={759--763},
doi={10.21437/Interspeech.2022-10463}
}