Contextualizing ASR Lattice Rescoring with Hybrid Pointer Network Language Model

TL;DR

This paper enhances ASR lattice rescoring by integrating video metadata through attention-based contextual vectors and a hybrid pointer network, improving recognition performance in social media videos.

Contribution

It introduces a novel hybrid pointer network approach and an attention-based method to incorporate video metadata into ASR lattice rescoring, which was not previously explored.

Findings

Both methods improve ASR performance by leveraging video metadata.

The hybrid pointer network explicitly models word probabilities from metadata.

Experimental results show significant performance gains.

Abstract

Videos uploaded on social media are often accompanied with textual descriptions. In building automatic speech recognition (ASR) systems for videos, we can exploit the contextual information provided by such video metadata. In this paper, we explore ASR lattice rescoring by selectively attending to the video descriptions. We first use an attention based method to extract contextual vector representations of video metadata, and use these representations as part of the inputs to a neural language model during lattice rescoring. Secondly, we propose a hybrid pointer network approach to explicitly interpolate the word probabilities of the word occurrences in metadata. We perform experimental evaluations on both language modeling and ASR tasks, and demonstrate that both proposed methods provide performance improvements by selectively leveraging the video metadata.