Hidden Markov Model (HMM) Toolbox for Matlab

This toolbox supports inference and learning for HMMs with discrete outputs (dhmm's), Gaussian outputs (ghmm's), or mixtures of Gaussians output (mhmm's). The Gaussians can be full, diagonal, or spherical (isotropic). It also supports discrete inputs, as in a POMDP. The inference routines support filtering, smoothing, and fixed-lag smoothing.

For a more recent version of this toolkit, please see PMTK.

Download toolbox

How to use the HMM toolbox

Other packages for HMMs

• Matlab package for teaching HMMs for Mendelian genetics
• HTK3 from Cambridge University is open source C code for HMMs for speech recognition.
• Mathworks stats toolbox 4.1 contains some functions for discrete HMMs.
• Zoubin Ghahramani has matlab code which is very similar to mine (but doesn't handle mhmm's). He also has code for approximate (variational) inference in factorial HMMs.
• Matlab Speech processing toolbox
• More matlab speech processing routines
• GNU HMM toolbox implements the basic algorithms.

Recommended reading on HMMs

• A tutorial on Hidden Markov Models and selected applications in speech recognition, L. Rabiner, 1989, Proc. IEEE 77(2):257--286.
• What HMMs can do, Jeff Bilmes, U. Washington Tech Report, Feb 2002
• Markovian Models for Sequential Data, Y. Bengio, Neural Computing Surveys 2, 129--162, 1999.
• Probabilistic independence networks for hidden Markov models P. Smyth, D. Heckerman, and M. Jordan, Neural Computation , vol.9, no. 2, 227--269, 1997.
• Notes from Dan Ellis' class on speech processing at Columbia (2002). I snagged the excellent notes for lecture 10 and Lecture 11.
• Krstulovic, Doss & Bourlard's EPFL matlab lab manuals on HMMs here .
• Bibliography on HMMs (2001)
• Bookmarks on HMMs
• Machine Learning for Sequential Data: A Review. Tom Dietterich. In T. Caelli (Ed.) Lecture Notes in Computer Science (2002).
• Teaching Baum Welch using Excel spreadsheets, Jason Eisner