Token-Level Supervised Contrastive Learning for Punctuation Restoration

TL;DR

This paper introduces a token-level supervised contrastive learning approach to improve punctuation restoration in speech recognition, addressing data imbalance issues and achieving significant F1 score improvements.

Contribution

It proposes a novel token-level supervised contrastive learning method specifically designed for punctuation restoration tasks.

Findings

Up to 3.2% absolute F1 improvement on test set

Addresses data imbalance in punctuation prediction

Enhances embedding space separation of punctuation classes

Abstract

Punctuation is critical in understanding natural language text. Currently, most automatic speech recognition (ASR) systems do not generate punctuation, which affects the performance of downstream tasks, such as intent detection and slot filling. This gives rise to the need for punctuation restoration. Recent work in punctuation restoration heavily utilizes pre-trained language models without considering data imbalance when predicting punctuation classes. In this work, we address this problem by proposing a token-level supervised contrastive learning method that aims at maximizing the distance of representation of different punctuation marks in the embedding space. The result shows that training with token-level supervised contrastive learning obtains up to 3.2% absolute F1 improvement on the test set.