moin David,
I haven't implemented Baum-Welch, but there is a Viterbi implementation in the [gfsm] library (which has Pd wrappers), although I haven't ported the Viterbi stuff to Pd yet, since it's still pretty new. In general, HMMs can be represented as finite state machines (a la gfsm), but if you really want a "pure" HMM (all transitions with nonzero probability), using an FSM would be a waste of memory. If you're interested anyways, gfsm is here:
http://www.ling.uni-potsdam.de/~moocow/projects/gfsm
... and the pd wrappers are on sf-cvs in externals/moocow/gfsm
marmosets, Bryan
On 02/07/2006 07:00:05 PM, David Plans Casal main@davidcasal.com appears to have written:
Hello all,
Looking back at 2003 in the archives, I see:
Johannes Taelman Johannes.Taelman at rug.ac.be Sun Feb 9 16:02:55 CET 2003
On Sat, 8 Feb 2003, Mark Khemma wrote:
check pd/doc/2.control.examples/21.markov.chain.pl
That's a markov chain, but not a hidden markov model. I was looking for the forward backward, viterbi, and baum-welch algorithms. I'll probably write an external for it sooner or later. Ideas welcome.
j#|@
Did Johannes ever write that? Has anyone else implemented HMM Baum- Welch?
d
-- David Plans Casal Researcher, UEA Studios d.casal at uea dot ac dot uk http://www.davidcasal.com
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list