Improving Hybrid CTC/Attention End-to-end Speech Recognition with Pretrained Acoustic and Language Model

TL;DR

This paper introduces a novel hybrid CTC/attention end-to-end speech recognition model that leverages pretrained acoustic and language models, achieving significant CER improvements on AISHELL-1 by fully utilizing self-supervised pretraining.

Contribution

It proposes a Preformer architecture that integrates pretrained acoustic and language models into a hybrid CTC/attention framework, enabling better utilization of pretraining in sequence-to-sequence ASR.

Findings

Achieved 4.6% CER on AISHELL-1 test set.

Realized 27% relative CER reduction over baseline.

First to utilize both pretrained acoustic and language models in a S2S ASR system.

Abstract

Recently, self-supervised pretraining has achieved impressive results in end-to-end (E2E) automatic speech recognition (ASR). However, the dominant sequence-to-sequence (S2S) E2E model is still hard to fully utilize the self-supervised pre-training methods because its decoder is conditioned on acoustic representation thus cannot be pretrained separately. In this paper, we propose a pretrained Transformer (Preformer) S2S ASR architecture based on hybrid CTC/attention E2E models to fully utilize the pretrained acoustic models (AMs) and language models (LMs). In our framework, the encoder is initialized with a pretrained AM (wav2vec2.0). The Preformer leverages CTC as an auxiliary task during training and inference. Furthermore, we design a one-cross decoder (OCD), which relaxes the dependence on acoustic representations so that it can be initialized with pretrained LM (DistilGPT2). Experiments are conducted on the AISHELL-1 corpus and achieve a $4.6\%$ character error rate (CER) on the test set. Compared with our vanilla hybrid CTC/attention Transformer baseline, our proposed CTC/attention-based Preformer yields $27\%$ relative CER reduction. To the best of our knowledge, this is the first work to utilize both pretrained AM and LM in a S2S ASR system.