I have implemented HHMM using MUBU objects, and they are pretty good describing gestures. Right now, I'm recording only three different examples of a gesture using three hhmm objects in the patch, and then estimation a media from the three likelihood outputs. If I want to train using 100 examples of a gesture, do I also need to create 100 hhmm-objects to fit them with one gesture each? Is it possible to send many versions of the gesture to only one hhmm-object? If you want to ask me about the patch, it is just precisely the same as the hhmm example from MUBU package. The only difference is that I have nine outputs gesture estimations.