Efficient Modeling of ITD Based on Anthropometrics and KEMAR ITD Coefficients Using Deep Neural Networks

Binaural cues, such as interaural time difference (ITD) and interaural level difference (ILD), are crucial for spatial hearing, especially for localizing sound sources. Machine learning, especially deep neural networks (DNNs), has shown promise in predicting individual ITD for various acoustics-related applications, including sound localization and assistive listening technologies. DNNs are effective in modeling ITD due to its correlation with specific anthropometric parameters. In this study, a single ITD for each subject is first modeled using DNNs. This modeled ITD is then combined with KEMAR's ITD coefficients to simulate a new user's ITDs across 1,250 different azimuth and elevation positions. The proposed ITD model is evaluated against known analytical models and recent methods using the mean absolute error (MAE) as the primary metric.