This paper introduces a novel framework for tracking the TDOAs of multiple sources whose acoustic activities overlap in time. Assuming the number of sources to be known, multiple disjoint particle filters estimate the posterior kernel density of the propagation parameters. An approximated instantaneous kernel density is provided through the Generalized State Coherence Transform, which improves the source interference rejection across the dimensions using a frequency-normalized non-linearity. Results obtained from an experimental evaluation on synthetic data show that the proposed framework enables localization and tracking of bidimensional TDOAs for 7 competitive sources recorded under reverberant conditions and with only 3 microphones.
Multiple Source Tracking by Sequential Posterior Kernel Density Estimation Through GSCT
Brutti, Alessio;Nesta, Francesco
2011-01-01
Abstract
This paper introduces a novel framework for tracking the TDOAs of multiple sources whose acoustic activities overlap in time. Assuming the number of sources to be known, multiple disjoint particle filters estimate the posterior kernel density of the propagation parameters. An approximated instantaneous kernel density is provided through the Generalized State Coherence Transform, which improves the source interference rejection across the dimensions using a frequency-normalized non-linearity. Results obtained from an experimental evaluation on synthetic data show that the proposed framework enables localization and tracking of bidimensional TDOAs for 7 competitive sources recorded under reverberant conditions and with only 3 microphones.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.