Anthropometric Features Estimation Using Integrated Sensors on a Headphone for HRTF Personalization

Islam MT, Tashev I. Anthropometric Features Estimation Using Integrated Sensors on a Headphone for HRTF Personalization. In: AES Conference: 2020 AES International Conference on Audio for Virtual and Augmented Reality (August 2020). Audio Engineering Society; 2020. Paper 7. Available from: https://aes.org/publications/elibrary-page/?id=20870

@inproceedings{Islam2020_20870,
  author = {Islam, Md Tamzeed and Tashev, Ivan},
  title = {{Anthropometric Features Estimation Using Integrated Sensors on a Headphone for HRTF Personalization}},
  booktitle = {AES Conference: 2020 AES International Conference on Audio for Virtual and Augmented Reality (August 2020)},
  note = {Paper 7},
  year = {2020},
  month = aug,
  publisher = {Audio Engineering Society},
  url = {https://aes.org/publications/elibrary-page/?id=20870}
}

TY  - CPAPER
TI  - Anthropometric Features Estimation Using Integrated Sensors on a Headphone for HRTF Personalization
AU  - Islam, Md Tamzeed
AU  - Tashev,Ivan
T2  - AES Conference: 2020 AES International Conference on Audio for Virtual and Augmented Reality (August 2020)
M1  - Paper 7
PY  - 2020
DA  - 2020/08/06
UR  - https://aes.org/publications/elibrary-page/?id=20870
PB  - Audio Engineering Society
LA  - en
AB  - Personalization of HRTF is essential for spatial sound rendering, for which a possible solution is based on one or more anthropological measures of the subject. Measuring these anthropometrics seamlessly, accurately and reliably is still a challenge. In this paper, we propose a system for obtaining anthropometric measurements, suitable for HRTF personalization, directly from a high-end headphone. The proposed system is multimodal and leverages existing sensors to extract features related to listener’s head dimensions. We propose three signal processing methodologies for three modalities of sensors and a fusion algorithm to aggregate these extracted features for a robust anthropometry estimation. To verify the design we use a data set, collected from 35 subjects. The proposed algorithm achieves a low error (RMSE) of 0.58 - 1.21 cm for human anthropometry estimation.
ER  -