The INTERSPEECH 2020 Computational Paralinguistics Challenge: Elderly Emotion, Breathing & Masks

Abstract

The INTERSPEECH 2020 Computational Paralinguistics Challenge addresses three different problems for the first time in a research competition under well-defined conditions: In the Elderly Emotion Sub-Challenge, arousal and valence in the speech of elderly individuals have to be modelled as a 3-class problem; in the Breathing Sub-Challenge, breathing has to be assessed as a regression problem; and in the Mask Sub-Challenge, speech without and with a surgical mask has to be told apart. We describe the Sub-Challenges, baseline feature extraction, and classifiers based on the 'usual' COMPARE and BoAW features as well as deep unsupervised representation learning using the AUDEEP toolkit, and deep feature extraction from pre-trained CNNs using the DEEP SPECTRUM toolkit; in addition, we partially add deep end-to-end sequential modelling, and, for the first time in the challenge, linguistic analysis.

An Evaluation of Speech-Based Recognition of Emotional and Physiological Markers of Stress

“…These features have shown to be effective for a number of similar wellbeing related tasks (Kim et al, 2019;? ;Schuller et al, 2020), including detection of early stage dementia (Haider et al, 2019), and levels of anxiety (Baird et al, 2020). For the DEEPSPECTRUM features, we extract a 2,560 dimensional feature set of deep data-representations using the DEEPSPECTRUM toolkit (Amiriparian et al, 2017).…”

Towards sound based testing of COVID-19—Summary of the first Diagnostics of COVID-19 using Acoustics (DiCOVA) Challenge

“… Deshpande and Schuller (2021) explored the estimates of breathing patterns, obtained from different sound categories, for COVID-19 detection. Towards this, an encoder which predicts breathing pattern from speech signals was designed using a subset of UCL Speech Breath Monitoring (UCL-SBM) database ( Schuller et al, 2020 ). This pre-trained encoder is then used to predict the breathing patterns from breathing, vowel-[i], and counting sound categories, separately.…”

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

“…Thus, the first thing that comes into one’s mind might be the detection of speech under a cold [25] . In the ongoing ComParE 2020 challenge, the continuous assessment of breathing patterns is proposed [26] . Moreover, automatically recognizing speech under a pain symptom [27] , [28] could be useful for an early warning.…”

“…Apart from the aforementioned individual aspects, social effects by COVID-19 can trigger another issue, e.g., the monitoring, management, and evaluation of the social distancing and quarantine. The social isolation of elderly may generate a serious public mental health issue, which is discussed as an emotion recognition task included in this year’s ComParE challenge [26] . Speaker identification and counting could be used for monitoring the social distancing, which can be implemented easily via smartphones [40] .…”