Ways to speedup SVR training in scikitlearn

Hi

I am using the scikitlearn implementation of Nu-SVR. My problem (automatic phonetic segmentation for singing voice) has ~ 50k points with 36 features. Seems relatively small to me compared to the datasets I have been reading about. The problem is it takes a long time (couple of hours) to fit the NuSVR model to 60% of this data and coarse gridsearching for the parameters obviously takes even longer (a couple of days).

Maybe there is some basic fault in my code, since I am very new to this.

For anyone interested, this is the inspiration for the approach.

thanks!