Created by: tiagofrepereira2012
Our bob.learn.em package would be more complete with a HMM implementation, what do you think?
The options to approach this are the following:
- Implement it from scratch (We already have something for the emission probability distribution)
- We can try to bind the HTK library (http://htk.eng.cam.ac.uk/) or try to integrate this bind http://www.e-gymraeg.org/wispr/pyhtk.htm
- Try to integrate different software https://github.com/jmschrei/pomegranate or sklearn.
Worth the effort?