Streaming, fast and accurate on-device Inverse Text Normalization for Automatic Speech Recognition

TL;DR

This paper presents a streaming, lightweight, and accurate on-device Inverse Text Normalization system for ASR that combines a transformer tagger with category-specific WFSTs, enabling efficient deployment on embedded devices.

Contribution

It introduces a novel on-device ITN approach using a streaming transformer tagger and category-specific WFSTs, reducing size and runtime costs while maintaining accuracy.

Findings

Achieves equivalent accuracy to strong baselines.

Significantly smaller model size.

Retains customization capabilities.

Abstract

Automatic Speech Recognition (ASR) systems typically yield output in lexical form. However, humans prefer a written form output. To bridge this gap, ASR systems usually employ Inverse Text Normalization (ITN). In previous works, Weighted Finite State Transducers (WFST) have been employed to do ITN. WFSTs are nicely suited to this task but their size and run-time costs can make deployment on embedded applications challenging. In this paper, we describe the development of an on-device ITN system that is streaming, lightweight & accurate. At the core of our system is a streaming transformer tagger, that tags lexical tokens from ASR. The tag informs which ITN category might be applied, if at all. Following that, we apply an ITN-category-specific WFST, only on the tagged text, to reliably perform the ITN conversion. We show that the proposed ITN solution performs equivalent to strong baselines, while being significantly smaller in size and retaining customization capabilities.