openfst
« Back to VersTracker
Description:
Library for weighted finite-state transducers
Type: Formula  |  Tracked Since: Dec 28, 2025
Links: Homepage  |  formulae.brew.sh
Category: Developer tools
Tags: c-plus-plus library finite-state transducer nlp speech-processing
Install: brew install openfst
About:
OpenFST is a library for constructing, combining, optimizing, and searching weighted finite-state transducers (FSTs). It provides a general-purpose software platform for rapid prototyping and development of complex sequential and context-dependent models. Its main value is offering efficient algorithms for dynamic programming and graph manipulation, widely used in speech recognition and natural language processing.
Key Features:
  • Efficient composition, determinization, and minimization of transducers
  • Support for various weight semirings (tropical, log, probability)
  • Extensible architecture for custom arc types and algorithms
  • C++ library with bindings for Python and Java
Use Cases:
  • Speech recognition and text-to-speech systems
  • Natural language processing (tokenization, parsing, machine translation)
  • Information extraction and pattern matching in large text corpora
  • Bioinformatics sequence analysis
Alternatives:
  • Kaldi – Kaldi is a full speech recognition toolkit that uses OpenFST for its graph operations, making them complementary rather than direct alternatives.
  • Pynini – Pynini is a Python library for finite-state grammar compilation that is built on top of OpenFST, offering a higher-level interface.
Version History
Detected Version Rev Change Commit
Sep 11, 2025 7:40am 0 VERSION_BUMP e5805139