SRILM - an extensible language modeling toolkit

SRILM is an extensible language modeling toolkit consisting of C++ libraries, executable programs, and helper scripts, designed to support both the production of and experimentation with statistical language models for speech recognition and other applications. It is freely available for noncommercial purposes. The toolkit supports the creation and evaluation of a variety of language model types based on N-gram statistics, as well as several related tasks such as statistical tagging and the manipulation of N-best lists and word lattices.

The paper summarizes the toolkit's functionality and discusses its design and implementation, highlighting ease of rapid prototyping, reusability of components, and the ability to combine tools into larger workflows. By offering a flexible, freely available platform for N-gram language modeling, SRILM became a standard toolkit in the speech recognition and natural language processing communities for years after its release.