Comparing Face-Tracking Action Units with EMG During Speech

Authors

Abstract

Video face-tracking software such as OpenFace 2.2 can make inferences about facial muscle activation (Baltrušaitis et al. 2018). However, it is unclear what the accuracy of these inferences is based on the facial action units (FAUs) calculated by OpenFace 2.2 compared to corresponding muscle activity during speech A previous study investigated muscle activation when a smile and speech co-occur, focusing on the Zygomaticus Major (ZM) and Orbicularis Oris (OO) muscles (Liu et al. 2021), but only presented data for a single speaker and did not compare FAU and EMG results. The present study compares OpenFace 2.2 action units with surface electromyography (EMG) data during speech in order to assess the validity of these inferences about facial muscle activation. We compare ZM activity with the lip corner puller FAU intensity results and OO activity with the lip-tightener FAU intensity results from a dataset collected for the previously mentioned Liu et al. (2021) study. Data includes four speakers producing read speech in smile conditions. We will report the results of a study extracting and analyzing 4-second segments from EMG and FAU data, recorded before and after each utterance, using Python for efficient data processing and graphical comparison. Our analysis to date indicates a relative correspondence between EMG and FAU data indicating the possible reliability of video face-tracking software in research applications. Details will be reported.

Additional Files

Published

2024-05-23

How to Cite

1.
Nozadi H. Comparing Face-Tracking Action Units with EMG During Speech. Canadian Acoustics [Internet]. 2024 May 23 [cited 2024 Jul. 4];52(1). Available from: https://jcaa.caa-aca.ca/index.php/jcaa/article/view/4196