Weighted Finite-State Transducers: The Later Years

Date Presented: 06/23/2022
Speaker: Kyle Gorman, CUNY

Abstract:
While the “deep learning tsunami” defines the state of the art in speech and language processing, finite-state transducer grammars developed by linguists and engineers are still widely used in highly-multilingual settings, particularly for “front-end” speech applications. In this talk, I will first briefly review the current state of the OpenFst and OpenGrm finite-state transducer libraries. I will then discuss several recent innovations in the finite-state world. These include algorithms for inducing text normalization and grapheme-to-phoneme grammars from parallel data, heuristic optimization of arbitrary weighted transducers, and an algorithm for efficiently computing the single shortest string of a wider variety of non-deterministic weighted acceptors.

Speaker's Bio:
Kyle Gorman is an assistant professor of linguistics at the Graduate Center, City University of New York, and director of the master’s program in computational linguistics; he is also a software engineer in the speech and language algorithms group at Google. With Richard Sproat, he is the coauthor of Finite-State Text Processing (Morgan & Claypool, 2021) and the creator of Pynini, a finite-state text processing library for Python. He has also published on statistical methods for comparing computational models, text normalization, grapheme-to-phoneme conversion, and morphological analysis, as well as many topics in linguistic theory.