On 10/4/06, Yosvany Llerena Rodríguez <yosvanyllr(a)gmail.com> wrote: This has nothing to do with linux or even audio development... I shouldn't answer, but anyway: Implementing EM for GMM is not really difficult by reading some good paper on EM (the master thesis here is quite well done, actually, with extension using variationel Bayes methods for reinforcement learning:http://www.cs.ubc.ca/grads/resources/thesis/Nov03/Kejie_Bao.pdf#search=%22kejie%20bao%22 ). There is an implementation of EM for finite GMM in python available here: http://www.ar.media.kyoto-u.ac.jp/members/david/pyem.tar.gz which handles both diagonal and full covariance matrices. David