Abstract: A multi-level distortion measure (MLDM) is proposed as an objective to optimize deep neural network-based speech enhancement (SE) in both audio-only and audio-visual scenarios. The aim is to ...
Abstract: Vibration acceleration signals allow humans to perceive the surface characteristics of textures during tool-surface interactions. However, acquiring acceleration signals requires a ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results