We propose an aspect-model-based reference speaker weighting. The main idea of the approach is that the adapted model is a linear combination of a set of reference speakers like r...
We developa clustereddithering methodthatusesstochasticscreening and is able to perform an adaptive variation of the cluster size. This makes it possible to achieve optimal rendit...
We introduce an information theoretic method for nonparametric, nonlinear dimensionality reduction, based on the infinite cluster limit of rate distortion theory. By constraining...
In this paper we consider the long run average continuous control problem of piecewise-deterministic Markov processes (PDP's for short). The control variable acts on the jump ...
This paper introduces a novel statistical mixture model for probabilistic grouping of distributional histogram data. Adopting the Bayesian framework, we propose to perform anneale...