CITATION.cff

cff-version: 1.2.0
message: "If you use hmm-studio in your research, please cite it as below."
title: "hmm-studio: constrained Baum-Welch fits, topology authoring, and visualization for Hidden Markov Models"
type: software
authors:
  - given-names: Robin
    family-names: Denis
repository-code: "https://github.com/RoJLD/HMMstudio"
license: MIT
version: "1.1.0"
date-released: 2026-05-23
keywords:
  - hidden-markov-models
  - hmm
  - baum-welch
  - viterbi
  - constrained-estimation
  - topology
  - nhmm
  - sequence-modeling
abstract: >
  hmm-studio is a Python library and (forthcoming) web tool for authoring
  Hidden Markov Models with structurally constrained transition matrices
  (left-right, Bakis, lifecycle, branching topologies). It wraps a constrained
  Baum-Welch fit engine over hmmlearn and exposes a backend-agnostic interface
  so alternative engines (pomegranate, dynamax, pure-numpy) can be plugged in.
  It also implements non-homogeneous HMMs (NHMM) with covariate-dependent
  transitions.