Schedule as of Oct 11, 2022 - subject to change

Default Time Zone is EDT - Eastern Daylight Time

Back To Schedule
Thursday, October 20 • 10:00am - 10:20am
Personalized HRTF Estimation Based on One-to-Many Neural Network Architecture

Sign up or log in to save this to your schedule, view media, leave feedback and see who's attending!

In this paper, we propose a one-to-many neural network (NN)-based model to estimate  personalized head-related transfer function (HRTF). The proposed model comprises a feature representation module and an estimation module. The feature representation module provides a deep feature associated with anthropometric measurement data for a given sound direction. The estimation module is mainly constructed using a bi-directional long short-term memory layer with feature vectors from multiple directions, which results in estimated HRTFs simultaneously for all the multiple directions. The performance of the proposed personalized HRTF estimation method is evaluated using the Center for Image Processing and Integrated Computing (CIPIC) database. Experiments show that the proposed personalized HRTF estimation method reduces root mean square error and log spectral distance by 0.89 and 0.45 dB, respectively, compared to the conventional NN-based method.

Thursday October 20, 2022 10:00am - 10:20am EDT