Adaptive Sparse and Monotonic Attention for Transformer-based Automatic Speech Recognition

TL;DR

This paper introduces an adaptive sparse and monotonic attention mechanism for Transformer-based speech recognition, enhancing online processing and alignment modeling, leading to improved recognition performance.

Contribution

It proposes a novel integration of sparse and monotonic attention into Transformer ASR to address limitations in streaming and alignment modeling.

Findings

Improved recognition accuracy on benchmark datasets.

Effective modeling of monotonic alignments.

Enhanced attention mechanism for online ASR.

Abstract

The Transformer architecture model, based on self-attention and multi-head attention, has achieved remarkable success in offline end-to-end Automatic Speech Recognition (ASR). However, self-attention and multi-head attention cannot be easily applied for streaming or online ASR. For self-attention in Transformer ASR, the softmax normalization function-based attention mechanism makes it impossible to highlight important speech information. For multi-head attention in Transformer ASR, it is not easy to model monotonic alignments in different heads. To overcome these two limits, we integrate sparse attention and monotonic attention into Transformer-based ASR. The sparse mechanism introduces a learned sparsity scheme to enable each self-attention structure to fit the corresponding head better. The monotonic attention deploys regularization to prune redundant heads for the multi-head attention structure. The experiments show that our method can effectively improve the attention mechanism on widely used benchmarks of speech recognition.