Conversion of nearly monaural audio to 5.1-channel audio for portable multimedia devices

HRTF Estimation in the Wild

“…The works in [54,55] show positive results when selecting the HRTF with the closest anthropometric measurements to a new user. Other works, [8,9,15,51], use regression methods or deep learning to predict HRTF features from these anthropometric measurements, including works like [28,51,52] that use images of the ears along with anthropometric measurements. Anthropometric methods face the same challenges as 3D scan-based methods, as obtaining accurate measurements is difficult and can be time-consuming.…”