Consultation des notices

Signal processing aspects of the modeling of head-related transfer functions (HRTFs) are examinated, with application to the real-time mixing and synthesis of two-channel signals for headphone or loudspeaker reproduction of three-dimensional sound images. The merits of IIR and FIR implementations of the synthesis filters are discussed and efficient solutions are proposed, including time-varying implementations (allowing head-tracking or the simulation of moving sound sources). Further modeling of early reflections and the use of feedback delay networks for reproducing the later diffuse reverberation make it possible to include accurate binaural room effect synthesis in the simulation, without exceeding the capacity of recent programmable digital signal processors.