Remember the context! ASR slot error correction through memorization

TL;DR

This paper introduces a k-NN based method to improve slot error correction in ASR systems by leveraging a datastore of domain-specific entities, significantly reducing word error rates on rare and out-of-vocabulary words.

Contribution

It proposes integrating k-NN search with a pretrained transformer model to enhance recognition of domain-specific and rare words in ASR outputs.

Findings

7.4% relative WER improvement on rare words

9.8% relative WER improvement on OOV words

Effective across five different domains

Abstract

Accurate recognition of slot values such as domain specific words or named entities by automatic speech recognition (ASR) systems forms the core of the Goal-oriented Dialogue Systems. Although it is a critical step with direct impact on downstream tasks such as language understanding, many domain agnostic ASR systems tend to perform poorly on domain specific or long tail words. They are often supplemented with slot error correcting systems but it is often hard for any neural model to directly output such rare entity words. To address this problem, we propose k-nearest neighbor (k-NN) search that outputs domain-specific entities from an explicit datastore. We improve error correction rate by conveniently augmenting a pretrained joint phoneme and text based transformer sequence to sequence model with k-NN search during inference. We evaluate our proposed approach on five different domains containing long tail slot entities such as full names, airports, street names, cities, states. Our best performing error correction model shows a relative improvement of 7.4% in word error rate (WER) on rare word entities over the baseline and also achieves a relative WER improvement of 9.8% on an out of vocabulary (OOV) test set.