The INTERSPEECH 2011 speaker state challenge

Computer Audition for Fighting the SARS-CoV-2 Corona Crisis—Introducing the Multitask Speech Corpus for COVID-19

“…To make a fair comparison with the current benchmark and future studies based on MSC-COVID-19, we use the unweighted average recall (UAR) as the main evaluation metric (e.g., to optimize the models’ hyperparameters on the dev set). UAR takes the data imbalance characteristics into account [85] , which can avoid an overly optimistic evaluation by using the weighted average recall (WAR), i.e., accuracy. Its value is defined as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}\begin{equation*} \text {UAR}=\frac {\sum \nolimits _{i=1}^{N_{\mathrm{class}}}{\mathrm{ Recall}}_{i}}{N_{\mathrm{class}}}\tag{1}\end{equation*} \end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}${\mathrm{ Recall}}_{i}$ \end{document} and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$N_{\mathrm{class}}$ \end{document} are the Recall of the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{upgreek} \usepackage{mathrsfs} \setlength{\oddsidemargin}{-69pt} \begin{document} }{}$i$ \end{document} th class and the number of classes, respectively.…”

“…It is also found that COVID-19 patients should have a lack of appetite [29] , which can be detected via the eating behavior analysis while speaking [30] . Sleepiness assessment can be implemented in both a binary classification task [31] and a regression estimation task (with Karolinska sleepiness scale) [32] . Considering the high mortality risk among the elderly group (a slightly higher mortality rate in male individuals) [24] , age and gender information could be of interest to be identified by speech [33] , [34] .…”

Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Student Research Workshop

“…The OpenFace (Baltrušaitis et al, 2015(Baltrušaitis et al, , 2018 software package was used to extract 17 different Ekman and Friesen (1976) facial action units (FACs) defined by per video frame. Audeering's openS-MILE (Schuller et al, 2009;Eyben et al, 2010) toolkit was used to extract 34 different audio features, including pitch, intensity, speech rate, and MFCCs per frame. Finally, Amazon Transcribe Using the output of Amazon Transcribe, each participant's text was divided into a sequential list of tokens, where a token could be a spoken word or a period of silence.…”

Multimodal Modeling of Task-Mediated Confusion

“…The OpenFace (Baltrušaitis et al, 2015(Baltrušaitis et al, , 2018 software package was used to extract 17 different Ekman and Friesen (1976) facial action units (FACs) defined by per video frame. Audeering's openS-MILE (Schuller et al, 2009;Eyben et al, 2010) toolkit was used to extract 34 different audio features, including pitch, intensity, speech rate, and MFCCs per frame. Finally, Amazon Transcribe Using the output of Amazon Transcribe, each participant's text was divided into a sequential list of tokens, where a token could be a spoken word or a period of silence.…”