Neural-FST Class Language Model for End-to-End Speech Recognition

TL;DR

This paper introduces NFCLM, a novel speech recognition model combining neural networks and FSTs, improving accuracy and compactness for on-device applications.

Contribution

The paper presents a new framework that integrates neural language models with FSTs using a neural decider, enhancing performance and efficiency.

Findings

NFCLM outperforms NNLM by 15.8% relative WER reduction.

NFCLM matches traditional NNLM and FST fusion performance.

NFCLM is 12 times more compact, suitable for on-device use.

Abstract

We propose Neural-FST Class Language Model (NFCLM) for end-to-end speech recognition, a novel method that combines neural network language models (NNLMs) and finite state transducers (FSTs) in a mathematically consistent framework. Our method utilizes a background NNLM which models generic background text together with a collection of domain-specific entities modeled as individual FSTs. Each output token is generated by a mixture of these components; the mixture weights are estimated with a separately trained neural decider. We show that NFCLM significantly outperforms NNLM by 15.8% relative in terms of Word Error Rate. NFCLM achieves similar performance as traditional NNLM and FST shallow fusion while being less prone to overbiasing and 12 times more compact, making it more suitable for on-device usage.