Parametric control of convolution based room simulators

For several decades, reverberation processors have been based on feedback delay networks that provide an efficient way to control the distribution of early reflections and the statistical properties of room reverberation. Their algorithmic architecture also allows for a flexible control through different levels of descriptors, from low-level signal processing parameters (e.g. initial delay, echo density, reverberation decay, etc.) to high-level perceptual descriptors (e.g. source presence, envelopment, reverberance, etc.). Nowadays, most reverberation units rather opt for a convolution approach exploiting a collection of room impulse responses (RIR) measured in existing performance halls or other typical rooms. To retrieve a continuous control over important perceptual dimensions, such reverberators also provide means for tweaking the original impulse response. However the tuning is generally limited to the control of the time envelope of the RIR, whereas little attention is given to the spatial distribution. The objective of this study is to develop a signal-processing environment dedicated to the analysis and re-synthesis of directional room impulse responses (DRIRs) measured with higher-order spherical microphone arrays. The analysis step results in a space-time-frequency representation of the DRIR from which various acoustical or perceptual descriptors can be extracted. During the re-synthesis step, filtering a given time-space-frequency window of the DRIR enables the tuning of these descriptors without altering the microstructure of the original DRIR.