Contribution
First experiments on how to blend playback and recording rooms in headphone based reproduction of multichannel sound
* Presenting author
Abstract:
It has been shown that the plausible reproduction of sound sources in a room is possible if the rendering algorithm uses the most important cues for localization of audio sources. This includes fast and accurate 6DOF tracking of the listeners current and constantly changing position. Another important cue is a model of the listening environment which accurately represents the sound source and receiver relation through a binaural room impulse response. It has been found, that for the reproduced BRIRs, a simple room model, the direct sound, a few early reflections and a common late reverberation model is good enough. In addition, a good generalized HRTF is needed.Such systems enable the reproduction of multichannel loudspeaker systems over headphones, e.g. 9.1.6 with 16 loudspeakers, without the need to have that number of physical speakers available. For this setup, a basic problem arises:According to the room divergence effect, a model of the actual room and not of another recording environment should be used for the most plausible and externalized reproduction.The talk will explain this problem and show some first anecdotal evidence of how to blend the current and the remote room acoustics to achieve a convincing binaural reproduction.