Monotonic Gaussian regularization of attention for robust automatic speech recognition. (January 2023)

• Record Type: Journal Article
• Authors: Du, Yeqian; Wu, Minghui; Fang, Xin; Yang, Zhouwang
• Is Part Of: Computer speech & language. Volume 77(2023)
• Publication Date: 2023-01
• Subjects: Speech recognition -- Multi-head attention -- Monotonic alignment -- Gaussian distribution -- Regularization -- Alignment robustness; Speech processing systems -- Periodicals; Automatic speech recognition -- Periodicals; Computers -- Periodicals; Linguistics -- Periodicals; Speech-Language Pathology -- Periodicals; Traitement automatique de la parole -- Périodiques; Reconnaissance automatique de la parole -- Périodiques; Automatic speech recognition; Speech processing systems; Electronic journals; Periodicals; 006.454
• DOI: 10.1016/j.csl.2022.101405
• Languages: English
• ISSNs: 0885-2308

Abstract

Abstract: The Attention-based Encoder–Decoder (AED) models are one of the most popular models for Automatic Speech Recognition (ASR). However, instability can occur in AED with problems such as incorrect insertions or word repetitions due to the violation of the inherent monotonic alignment property. To address these problems, we propose a monotonic Gaussian regularization method to guide the attention training, where the guiding map is depicted as a sequence of Gaussian distributions with monotonically moving centers. Experiments show our method reduces the insertion error rate by a relative 7% on the HKUST dataset, relative 20% and 16% on two large industrial datasets, and a relative 21% on an out-of-domain test set. The overall Character Error Rates (CERs) are all reduced at the same time, indicating that the model's recognition ability is well maintained. Therefore, our proposed method improves model performance by enhancing monotonic alignment, and provides better robustness. Highlights: Introduction of monotonic alignment property to improve model robustness. A new Gaussian regularization method to achieve monotonicity. Comparison with previous work based on error rates and attention plots. Tests under various conditions show a significant reduction in insertion errors and improvements in CER. Attention visualization and decoding examples illustrate the effect of the monotonicity constraint.