Supplementary site for DAFX 2017 paper

CONSTRAINED POLE OPTIMIZATION FOR MODAL REVERBERATION

Authors
Esteban Maestre, Jonathan S. Abel, Julius O. Smith, Gary P. Scavone

Abstract
The problem of designing a modal reverberator to match a measured room impulse response is considered. The modal reverberator architecture expresses a room impulse response as a parallel combination of resonant filters, with the pole locations determined by the room resonances and decay rates, and the zeros by the source and listener positions. Our method first estimates the pole positions in a frequency-domain process involving a series of constrained pole position optimizations in overlapping frequency bands. With the pole locations in hand, the zeros are fit to the measured impulse response using least squares. Example optimizations for a medium-sized room show a good match between the measured and modeled room responses.

Sound examples

Below we provide several impulse response modeling examples, followed by two demo uses. In all models, which were obtained from the same target measurement (fs = 48kHz, 1.25 seconds), M poles were estimated via B = 200 bands in between 30 Hz and 20000 Hz, with a band overlap factor delta = 1.0. To give an idea, the computational cost of models with M = 1800 poles is around 10 times smaller than that of convolving with the target measurement. Moreover, the cost can be further reduced, as the modal structure can be seamlessly implemented via parallel architectures.


IR MODELING DEMO #1 DEMO #2
Target measurement

M = 400
(target IR, modal IR)

M = 800
(target IR, modal IR)

M = 1200
(target IR, modal IR)

M = 1800
(target IR, modal IR)
Dry signal

M = 400
(full convolution, modal reverb)

M = 800
(full convolution, modal reverb)

M = 1200
(full convolution, modal reverb)

M = 1800
(full convolution, modal reverb)
Dry signal

M = 400
(full convolution, modal reverb)

M = 800
(full convolution, modal reverb)

M = 1200
(full convolution, modal reverb)

M = 1800
(full convolution, modal reverb)