PyChain: A Fully Parallelized PyTorch Implementation of LF-MMI for End-to-End ASR

TL;DR

PyChain is a flexible, fully parallelized PyTorch implementation of LF-MMI training for end-to-end ASR, enabling efficient GPU training and integration with existing tools, achieving competitive results on WSJ.

Contribution

It introduces a lightweight, flexible PyTorch-based LF-MMI training toolkit that supports GPU acceleration and can be integrated into various ASR projects.

Findings

Achieves competitive results on WSJ dataset.

Supports full GPU training on numerator/denominator graphs.

Demonstrates flexibility and efficiency in ASR training.

Abstract

We present PyChain, a fully parallelized PyTorch implementation of end-to-end lattice-free maximum mutual information (LF-MMI) training for the so-called \emph{chain models} in the Kaldi automatic speech recognition (ASR) toolkit. Unlike other PyTorch and Kaldi based ASR toolkits, PyChain is designed to be as flexible and light-weight as possible so that it can be easily plugged into new ASR projects, or other existing PyTorch-based ASR tools, as exemplified respectively by a new project PyChain-example, and Espresso, an existing end-to-end ASR toolkit. PyChain's efficiency and flexibility is demonstrated through such novel features as full GPU training on numerator/denominator graphs, and support for unequal length sequences. Experiments on the WSJ dataset show that with simple neural networks and commonly used machine learning techniques, PyChain can achieve competitive results that are comparable to Kaldi and better than other end-to-end ASR systems.