Improved Mask-CTC for Non-Autoregressive End-to-End ASR

• Main Authors: Higuchi, Yosuke, Inaguma, Hirofumi, Watanabe, Shinji, Ogawa, Tetsuji, Kobayashi, Tetsunori
• Format: Conference Proceeding
• Language: English
• Subjects: Conferences; connectionist temporal classification; Decoding; Degradation; end-to-end speech recognition; end-to-end speech translation; Network architecture; Non-autoregressive sequence generation; Predictive models; Signal processing; Training
• ISSN: 2379-190X
• DOI: 10.1109/ICASSP39728.2021.9414198

For real-world deployment of automatic speech recognition (ASR), the system is desired to be capable of fast inference while relieving the requirement of computational resources. The recently proposed end-to-end ASR system based on mask-predict with connectionist temporal classification (CTC), Mask-CTC, fulfills this demand by generating tokens in a non-autoregressive fashion. While Mask-CTC achieves remarkably fast inference speed, its recognition performance falls behind that of conventional autoregressive (AR) systems. To boost the performance of Mask-CTC, we first propose to enhance the encoder network architecture by employing a recently proposed architecture called Conformer. Next, we propose new training and decoding methods by introducing auxiliary objective to predict the length of a partial target sequence, which allows the model to delete or insert tokens during inference. Experimental results on different ASR tasks show that the proposed approaches improve Mask-CTC significantly, outperforming a standard CTC model (15.5% → 9.1% WER on WSJ). Moreover, Mask-CTC now achieves competitive results to AR models with no degradation of inference speed (< 0.1 RTF using CPU). We also show a potential application of Mask-CTC to end-to-end speech translation.