# Modeling Acoustic-Prosodic Cues for Word Importance Prediction in Spoken Dialogues

## TL;DR

This paper explores how acoustic-prosodic cues in spoken dialogues can predict the importance of words, aiding transcription systems especially when dealing with imperfect speech recognition outputs.

## Contribution

It introduces neural architectures that utilize acoustic features for word importance prediction, performing competitively with text-based models and benefiting ASR systems.

## Key findings

- Neural models using acoustic cues achieve competitive accuracy.
- Acoustic-based prediction improves robustness on imperfect ASR output.
- The approach benefits real-time captioning for Deaf and Hard of Hearing users.

## Abstract

Prosodic cues in conversational speech aid listeners in discerning a message. We investigate whether acoustic cues in spoken dialogue can be used to identify the importance of individual words to the meaning of a conversation turn. Individuals who are Deaf and Hard of Hearing often rely on real-time captions in live meetings. Word error rate, a traditional metric for evaluating automatic speech recognition, fails to capture that some words are more important for a system to transcribe correctly than others. We present and evaluate neural architectures that use acoustic features for 3-class word importance prediction. Our model performs competitively against state-of-the-art text-based word-importance prediction models, and it demonstrates particular benefits when operating on imperfect ASR output.