ASR Error Correction with Constrained Decoding on Operation Prediction

TL;DR

This paper introduces a constrained decoding method for ASR error correction that predicts correction operations to reduce latency and improve inference speed without sacrificing accuracy, supported by experiments on public datasets.

Contribution

It proposes a novel operation prediction-based correction method with a predictor module, significantly reducing decoding latency while maintaining accuracy, and releases a benchmark dataset for ASR correction.

Findings

Inference speed increased by 3.4 to 5.7 times.

WER reduced by up to 1.69%.

Effective on multiple datasets.

Abstract

Error correction techniques remain effective to refine outputs from automatic speech recognition (ASR) models. Existing end-to-end error correction methods based on an encoder-decoder architecture process all tokens in the decoding phase, creating undesirable latency. In this paper, we propose an ASR error correction method utilizing the predictions of correction operations. More specifically, we construct a predictor between the encoder and the decoder to learn if a token should be kept ("K"), deleted ("D"), or changed ("C") to restrict decoding to only part of the input sequence embeddings (the "C" tokens) for fast inference. Experiments on three public datasets demonstrate the effectiveness of the proposed approach in reducing the latency of the decoding process in ASR correction. It enhances the inference speed by at least three times (3.4 and 5.7 times) while maintaining the same level of accuracy (with WER reductions of 0.53% and 1.69% respectively) for our two proposed models compared to a solid encoder-decoder baseline. In the meantime, we produce and release a benchmark dataset contributing to the ASR error correction community to foster research along this line.