The aim of sound field reproduction is to reproduce a sound field in an extended area using loudspeaker arrays. Both the spatial and the spectral features are required to reproduce the original sound field. The spatial information (i.e. relative amplitudes and phases along with time differences between microphones' signals) can be retrieved from microphone array recordings of the original sound field. However, the original spectral features could also be modified by the spectral features recorded by a supplementary external microphone. The resulting modified sound field would then inherit its general spectral magnitude from the external microphone and its spatial characteristics from the original microphone array recording. Such a possibility opens an interesting avenue for virtual acoustics. To this end, cross-synthesis, a common method with practical applications in music technology and sound synthesis mostly applied to monophonic sounds, could be adapted to virtually modify the spectral envelop without affecting the spatial features of the original microphone array recordings. To do so, the spectral envelop of a reference microphone, part of the microphone array, is extracted using linear prediction and used to whiten the signals of the entire microphone array. Finally, using an external microphone, it is possible to colorize the microphone array recording with respect to the external microphone spectral envelop. Theoretical results based on real recordings are reported to illustrate the method.
