Graph Neural Networks for Contextual ASR with the Tree-Constrained Pointer Generator

TL;DR

This paper introduces a novel GNN-based method for end-to-end contextual ASR, significantly improving recognition accuracy for rare and unseen words by leveraging tree-constrained pointer generators.

Contribution

It proposes a new GNN encoding approach for contextual ASR that enhances prediction of biasing words, demonstrating substantial WER reductions on standard datasets.

Findings

Achieved over 60% WER reduction for rare and unseen words.

GNN encodings improve lookahead and prediction accuracy in ASR decoding.

Effective combination of GCN and GraphSAGE structures enhances performance.

Abstract

The incorporation of biasing words obtained through contextual knowledge is of paramount importance in automatic speech recognition (ASR) applications. This paper proposes an innovative method for achieving end-to-end contextual ASR using graph neural network (GNN) encodings based on the tree-constrained pointer generator method. GNN node encodings facilitate lookahead for future word pieces in the process of ASR decoding at each tree node by incorporating information about all word pieces on the tree branches rooted from it. This results in a more precise prediction of the generation probability of the biasing words. The study explores three GNN encoding techniques, namely tree recursive neural networks, graph convolutional network (GCN), and GraphSAGE, along with different combinations of the complementary GCN and GraphSAGE structures. The performance of the systems was evaluated using the Librispeech and AMI corpus, following the visual-grounded contextual ASR pipeline. The findings indicate that using GNN encodings achieved consistent and significant reductions in word error rate (WER), particularly for words that are rare or have not been seen during the training process. Notably, the most effective combination of GNN encodings obtained more than 60% WER reduction for rare and unseen words compared to standard end-to-end systems.