DFSMN-SAN With Persistent Memory Model For Automatic Speech Recognition

Abstract

Self-attention networks (SAN) have been introduced into automatic speech recognition (ASR) and achieved state-of-the-art performance owing to its superior ability in capturing long term dependency. One of the key ingredients is the self-attention mechanism which can be effectively performed on the whole utterance level. In this paper, we try to investigate whether even more information beyond the whole utterance level can be exploited and beneficial. We propose to apply self-attention layer with augmented memory to ASR. Specifically, we first propose a variant model architecture which combines deep feed-forward sequential memory network (DFSMN) with self-attention layers to form a better baseline model compared with a purely self-attention network. Then, we propose and compare two kinds of additional memory structures added into self-attention layers. Experiments on large-scale LVCSR tasks show that on four individual test sets, the DFSMN-SAN architecture outperforms vanilla SAN encode

Stats

• citations5
• S2 citations—
• github stars0
• HF likes0
• heat score5.84
• arxiv keyyou2019dfsmn

Related papers

• Simplified Self-attention For Transformer-based End-to-end Speech Recognition (2020)10.61
• Speech-xlnet: Unsupervised Acoustic Model Pretraining For Self-attention Networks (2019)10.35
• Unidirectional Memory-self-attention Transducer For Online Speech Recognition (2021)3.58
• Self-attention Networks For Connectionist Temporal Classification In Speech Recognition (2019)14.55
• A Novel Pyramidal-fsmn Architecture With Lattice-free MMI For Speech Recognition (2018)0.00
• Deep Feed-forward Sequential Memory Networks For Speech Synthesis (2018)5.84
• ASAPP-ASR: Multistream CNN And Self-attentive SRU For SOTA Speech Recognition (2020)9.03
• Streaming Chunk-aware Multihead Attention For Online End-to-end Speech Recognition (2020)8.60