I'm working on an alignment algorithm using LAMP HMM library. This library supports Gaussian probability distribution but it does not seem to support Gaussian Mixture Model.

What I want is, to input continuous observations into multivariate Gaussians for training and alignment purposes.

It may be easy if I extend the current Gaussian model to support multiple Gaussians but I don't understand what codes to change and what codes to type as I'm a beginner to this topic.

Does anyone know of websites for learning such concepts? Or can anyone show some pointers on how to do so?


  • $\begingroup$ You probably want to contact the developers of that library. $\endgroup$ – bayerj Jan 9 '12 at 6:27
  • $\begingroup$ It would make more sense to use the latent variable representation of a mixture and to add those latent variables as a secondary HMM with an independence constraint... $\endgroup$ – Xi'an Jan 9 '12 at 7:39

I have implemented Multivariate Gaussian Mixture after I've read through this article here

Gaussian Mixture Models and Expectation-Maximization


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.