Localization of Virtual Sounds in Dynamic Listening Using Sparse HRTFs