Skip to content
conftrace_
2020 EMNLP EMNLP 2020

The importance of fillers for text representations of speech transcripts

Abstract

AbstractWhile being an essential component of spoken language, fillers (e.g. “um” or “uh”) often remain overlooked in Spoken Language Understanding (SLU) tasks. We explore the possibility of representing them with deep contextualised embeddings, showing improvements on modelling spoken language and two downstream tasks — predicting a speaker’s stance and expressed confidence.

🌉 Interdisciplinary Bridge - Artificial Intelligence and Natural Language Processing and Speech & Audio

Authors

Tanvi Dinkar, Pierre Colombo, Matthieu Labeau, Chloé Clavel

Topics

Natural Language Processing > Resources & Methods > Text Representation Speech & Audio > Analysis > Speech Analysis Deep Learning > Learning Types > Representation Learning Artificial Intelligence > Core AI > Natural Language Processing

Keywords

text representation spoken language understanding contextual embedding speech transcript stance prediction filler word
Download PDF

More from EMNLP 2020

All EMNLP 2020 papers →
Active Learning for BERT: An Empirical Study
Transformer Based Multi-Source Domain Adaptation
Vector-Vector-Matrix Architecture: A Novel Hardware-Aware Framework for Low-Latency Inference in NLP Applications
The role of context in neural pitch accent detection in English
VolTAGE: Volatility Forecasting via Text Audio Fusion with Graph Convolution Networks for Earnings Calls
Effectively pretraining a speech translation decoder with Machine Translation data