World Voice Day was founded in the year 2002 and has been supported by the American Academy of Otolaryngology‒Head and Neck Surgery (AAO-HNS) ever since. The (AAO-HNS) represents about 12,000 otolaryngologists (head and neck surgeons) as it is the world’s largest organization for head and neck specialists. Each year, on the 16. April voice health professionals and surgeons aim to show the great value of the voice in people’s daily lives, encourage and inform about the prevention of voice issues as well as to highlight current research, applications and functions of the voice. Voice based research is applied in various disciplines like biology, phonetics, psychology, speech pathology and medicine.
ki:elements are pioneers in speech biomarkers and we use this day to show how to use the human voice as an innovative instrument for early detection of mental disease and stress by presenting this paper: “Measuring Stress in Health Professionals over the Phone using Automatic Speech Analysis during COVID-19 Pandemic” (König, Riviere, Linz et al., 2020). The study’s aim is to measure different stress levels in health professionals working during the COVID 19 pandemic. Within a 15 minute phone call, participants were asked to complete a short positive/negative and neutral storytelling task, which was used to analyse their stress levels based on their speech behavior. In the course of this, the study collected 89 speech samples of healthcare professionals of which various voice features were extracted and compared with classical stress measures via standard questionnaires for example.
Psychological stress manifests itself in different ways in the human body, for example as an increased muscle tension, increased breathing rate and voice pitch, which is one of the most common reported findings in studies examining speech under stress.These voice features affect the vocal production and the rate of vocal fold vibration. Due to stress, the vocal fold vibration has to stretch more and becomes more tense leading to a higher subglottal pressure1, vocal intensity and speech prosody. (cf. Giddens et al., 2013, Pisanski e al., 2016, Kirchhuebel et al., 2011) To measure stress in clinical practice, there are various scales and questionnaires such as the MSA (motivation, stress, affect) questionnaire or the the Perceived Stress Questionnaire (PSQ) – both of them were used in the study.
Automatic speech analysis enabled the extraction of voice features during the phone call using a software to analyse the voice with respect to acoustic and semantic parameters like the prosodic2, formant3, source4 and temporal5 characteristics. The study was able to show the correlation between speech characteristics and stress levels in both genders; mainly formant and prosodic features appeared to be sensitive to stress. For male and female participants, findings from the positive tasks yielded the most accurate prediction results of the stress levels. The combination of automatic speech analysis and intervention strategies may lead to a prevention of burnout as well as the development of depression or anxiety.
Another example where the automatic analysis of voice and speech can be used is as a prescreening tool to extract suitable participants from less suitable ones for example for clinical trials, read more here. There are currently different research fields where voice is used in order to prevent or detect chronic disease, for example a preventive program for University Lecturers’ vocal health based on acoustic signaling to monitor or decrease the risk of developing a voice disorder (cf. Paniagua et a., 2020).
1 the air pressure in the lungs, leading to development of the human voice
2 pitch and intonation
3 spectrum and acoustic resonance of the vocal tract
4 source of voice production
5 proportion of speech