views:

439

answers:

3

Hi,

I want to get started on HMM's, but don't know how to go about it. Can people here, give me some basic pointers, where to look?

More than just the theory, I like to do a lot of hands-on. So, would prefer resources, where I can write small code snippets to check my learning, rather than just dry text.

Will be hoping to see some replies.

cheers, R

+2  A: 

Have you tried: Russel and Norvig's Artificial Intelligence: A Modern Approach.

I realise that this is heavy on theory, but it also contains useful code samples that can be used to help your learning.

You can also check out: http://www.kanungo.com/software/software.html for a c-implementation of a HMM

TK
+2  A: 

Check out the Wikipedia article on HMMs: they have a pretty solid example after all the theory stuff. If you want to get some practice on it, Ruby Quiz has some great Markov model implementations that you can try changing to be HMMs.

Chris Bunch
+1  A: 

In our research lab, we generally use the HMM Toolkit to get started with HMM modelling. Unfortunately it has some licensing restrictions on redistribution (basically you can't redistribute the software, but you can redistribute models you've trained with it), but it may be useful to get started on learning how they work. The HTK Book provided with the HMM Toolkit is also a pretty comprehensive reference on HMM design.

If you want to get some data the may be useful for training HMMs, have a look at the VoxForge project, where you will also find some links to open source speech recognition systems that may be useful in getting your feet wet.

David Dean