There exist several options available for Gaussian Mixture Models in Python instead of using the sklearn library. Three at least :
PyMix - http://www.pymix.org/pymix/index.php Tools for mixture modeling
PyEM - http://www.ar.media.kyoto-u.ac.jp/members/david/softwares/em/ which is part of the Scipy toolbox and seems to focus on GMMs Update: Now known as sklearn.mixture .
PyPR - http://pypr.sourceforge.net/ pattern recognition and related tools including GMMs
I do not know the time computation of those method but I think they are pretty equivalent.
You can also look at https://github.com/zxytim/fast-gmm but there is no update for 2 years...