Using Ragel for Hidden Markov Model generation

[edward... at gmail.com]

Hi everyone,

I'm interested in using Ragel as the core of a Hidden Markov Model
(HMM) engine.

As I understand it, a HMM is really just a non-deterministic finite
state machine with probabilities that dictate the emissions and
transition probabilities. So instead of there being a completely random
choice between two epsilon transitions (for example), there's a
probability for each path.

Would it be a big deal to alter Ragel to support probabilities in state
transitions?

Thanks!

Edward