COVID-19 Patient Detection from Telephone Quality Speech Data

TL;DR

This study explores detecting COVID-19 from telephone-quality speech data using speaker recognition techniques, achieving high accuracy with SVM classifiers on a small dataset, and identifying specific phonemes that differentiate COVID-19 speech.

Contribution

It introduces a novel approach applying speaker recognition features to COVID-19 detection from speech, demonstrating promising results on YouTube data and highlighting phonemes most indicative of the disease.

Findings

SVM classifier achieved 88.6% accuracy and 92.7% F1-score.

Certain phonemes like nasals, stops, and mid vowels are more discriminative.

Method shows potential for non-invasive COVID-19 screening from speech.

Abstract

In this paper, we try to investigate the presence of cues about the COVID-19 disease in the speech data. We use an approach that is similar to speaker recognition. Each sentence is represented as super vectors of short term Mel filter bank features for each phoneme. These features are used to learn a two-class classifier to separate the COVID-19 speech from normal. Experiments on a small dataset collected from YouTube videos show that an SVM classifier on this dataset is able to achieve an accuracy of 88.6% and an F1-Score of 92.7%. Further investigation reveals that some phone classes, such as nasals, stops, and mid vowels can distinguish the two classes better than the others.