3D binaural sound reproduction using a virtual ambisonic approach

Convincing binaural sound reproduction via headphones requires filtering the virtual sound source signals with head related transfer functions (HRTFs). Furthermore, humans are able to improve their localization capabilities by small unconscious head movements. Therefore it is important to incorporate head-tracking. This yields the problem of high-quality, time-varying interpolation between different HRTFs. A further improvement of human localization accuracy can be done by considering room simulation yielding a huge amount of virtual sound sources. To increase the computational efficiency of the proposed system, a virtual ambisonic approach is used, that result in a bank of time-invariant HRTF filter independent of the number of sources to encode.