Robust ITD error estimation for crosstalk cancellation systems with a microphone-based head-tracker

Previously, we proposed a crosstalk cancellation system with a microphone-based head-tracker. The head-tracker estimates the orientation of the head using the interaural time difference (ITD) error between the desired binaural signal and the signals at the microphones, which are placed near the listener's ears. In the presence of noise and multiple virtual sound sources, simple cross-correlation based techniques will fail to find the true ITD error value. In this study, we propose a method to estimate the ITD error from multiple competing sources in a noisy environment. We use relative transfer functions (RTFs) to estimate the ITD error in subbands and apply a weighting function based not only on the coherence, but also on the magnitude of the RTFs in each subband. Experiments show that in the presence of multiple sources and noise, the proposed weighting function yields better results compared to a weighting function that depends only on the coherence.