The Objective and Subjective Sleepiness Voice Corpora

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 6523–6531 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 6523 The Objective and Subjective Sleepiness Voice Corpora Vincent P. Martin1, Jean-Luc Rouas1, Jean-Arthur Micoulaud Franchi2, Pierre Philip2 1LaBRI - CNRS - UMR 5800 - Univ. Bordeaux - Bordeaux INP - F-33400 Talence (France) 2SANPSY - CNRS - USR 3413 - Univ. Bordeaux - CHU Pellegrin - F-33000 Bordeaux (France) {vincent.martin, rouas}@labri.fr, {jean-arthur.micoulaud-franchi, pierre.philip}@u-bordeaux.fr Abstract Following patients with chronic sleep disorders involves multiple appointments between doctors and patients which often results in episodic follow-ups with unevenly spaced interviews. Speech technologies and virtual doctors can help improve this follow-up. However, there are still some challenges to overcome: sleepiness measurements are diverse and are not always correlated, and most past research focused on detecting instantaneous sleepiness levels of healthy sleep-deprived subjects. This article presents a large database to assess the sleepiness level of highly phenotyped patients that complain from excessive daytime sleepiness. Based on the Multiple Sleep Latency Test, it differs from existing databases by multiple aspects. First, it is composed of recordings from patients suffering from excessive daytime sleepiness instead of sleep deprived healthy subjects. Second, it incites the subjects to sleep contrary to existing stressing sleepiness deprivation experimental paradigms. Third, the sleepiness level of the patients is evaluated with different temporal granularities - long term sleepiness and short term sleepiness - and both objective and subjective sleepiness measures are collected. Finally, it relies on the recordings of 94 highly phenotyped patients, allowing to unravel the influences of different physical factors (age, sex, weight, ... ) on voice. Keywords: Sleepiness measurement and detection, Innovative healthcare, Read speech 1. Introduction 1.1. Motivations One of the major challenges for diagnosing and treat- ing neuro-psychiatric pathologies is symptom quantifica- tion and follow-up of chronic patients in order to adapt treatment and measure early relapses. Such an ecological monitoring is possible thanks to connected medical devices (measuring for instance weight, blood pressure or physical activities) but crucial information about how the patients report clinical symptoms like fatigue or sleepiness are dif- ficult to measure. Regular in-person appointments between doctors and patients are useful but miss a large part of variability of symptoms at home in response to treatment. Furthermore, the growing number of patients increases the queuing time and often results in episodic follow-ups with unevenly spaced interviews. Apart from the clinical interviews, it is nonetheless possi- ble to measure some symptoms (e.g. sadness or sleepiness) with a range of behavioural analysis techniques: looking at eye movements and examining verbal expressions or body movements (Poursadeghiyan et al., 2018; Khan and Man- soor, 2008). Thanks to recent advances in speech process- ing, it seems now possible to detect precise cues in voice allowing to characterise the state of a speaker. This could potentially allow to measure the level of sleepiness, fatigue or sadness (Cummins et al., 2018). This method has mul- tiple advantages as recording voice data is not …

Content truncated to 500 words