Using Text Injection to Improve Recognition of Personal Identifiers in Speech

TL;DR

This paper proposes a text-injection method to enhance the recognition accuracy of personal identifiers like names and dates in speech recognition systems, balancing privacy concerns with improved performance.

Contribution

It introduces a novel text-injection technique that incorporates fake PII data into training to improve recognition without compromising privacy.

Findings

Significant improvement in Recall of Names and Dates in medical notes

Enhanced Character Error Rate and Sentence Accuracy for digit sequences

Overall WER is improved through the proposed method

Abstract

Accurate recognition of specific categories, such as persons' names, dates or other identifiers is critical in many Automatic Speech Recognition (ASR) applications. As these categories represent personal information, ethical use of this data including collection, transcription, training and evaluation demands special care. One way of ensuring the security and privacy of individuals is to redact or eliminate Personally Identifiable Information (PII) from collection altogether. However, this results in ASR models that tend to have lower recognition accuracy of these categories. We use text-injection to improve the recognition of PII categories by including fake textual substitutes of PII categories in the training data using a text injection method. We demonstrate substantial improvement to Recall of Names and Dates in medical notes while improving overall WER. For alphanumeric digit sequences we show improvements to Character Error Rate and Sentence Accuracy.