Excessive sleepiness is a significant public health issue and a critical personal health indicator associated with various disorders. Given its high prevalence in the general population, clinicians need tools to regularly measure patients’ sleepiness levels in natural settings, such as automatic speech analysis. In this article, we introduce the SOMVOICE corpus, the first French corpus containing read-speech recordings from the same participants either after a normal night or after a night of total sleep deprivation. Participants were included according to strict inclusion and exclusion criteria based on both medical characteristics and reading proficiency. The recordings were labelled with both objective and subjective measures of sleepiness, as well as fatigue and anxiety. After introducing the data-collection methodology, we use linear mixed models to conduct a preliminary investigation of the effect of total sleep deprivation on the collected sleepiness-related measures and on participants’ reading behaviour. Doing so, we found that sleep deprivation strongly influences objective and subjective sleepiness measurements as well as fatigue self-reports, but has a lesser effect on anxiety. Regarding reading behaviour, sleep deprivation is associated with a lower speech rate (duration of the recordings and phoneme rate) and more pauses (number of pauses and pause ratio)

Excessive Daytime Sleepiness (EDS) is associated with several diseases and therefore negatively affects the daily life of impacted people. Its diagnosis and follow-up are difficult because they require testing at the hospital for one full day. Monitoring patients regularly in ecological conditions may be done through speech analysis. Although several corpora containing speech from sleepy subjects exist, they do not suit ecological requirements regarding either the device used for recording or the speech elicitation tasks. In this paper, we introduce the Medispeech corpus containing reading, daily-life semi-spontaneous, and medically-oriented spontaneous tasks. Fifty-nine French subjects were recorded with both a professional-quality microphone and a smartphone using a dedicated application, resulting in 1,729 recordings for a total duration of 21 hours. Their EDS diagnosis was assessed by both a physiological objective measurement (mean sleep latency measured during a clinical test) and a subjective questionnaire (Karolinska Sleepiness Scale). Phenotyping of subjects is assured by collecting socio-demographic and medical data related to diverse dimensions of sleepiness, comorbidities, and addictions. Finally, we analyse the validity of our data collection protocol by measuring the effective duration of speech (after discarding pauses) and assessing its links with the collected subjects’ characteristics.

La Somnolence Diurne Excessive affecte négativement les individus et est un problème de santé publique. L’analyse de la parole pourrait aider les cliniciens à la surveiller. Nous nous sommes concentrés sur la détection du schwa /@/ et avons trouvé un lien entre le nombre d’occurrences annoté manuellement et le niveau de somnolence des patients hypersomnolents d’un sous-ensemble du corpus TILE. Dans un second temps, afin de pouvoir généraliser ces résultats à l’intégralité du corpus, nous avons conçu un système de détection des schwas, robuste à la somnolence. Dans un troisième temps, nous avons étendu notre analyse à deux autres phonèmes supplémentaire /ø/ et /oe/. Nous avons ainsi observé une relation significative entre /ø/ et la combinaison des trois phonèmes et la somnolence subjective à court terme.

Le suivi des patients souffrant de maladies neuro-psychiatriques chroniques peut être amélioré grâce à la détection de la somnolence dans la voix. Cet article s’inspire des systèmes état-de-l’art en détection de la somnolence dans la voix pour le cas particulier de patients atteints de Somnolence Diurne Excessive (SDE). Pour cela, nous basons notre étude sur un nouveau corpus, le corpus TILE. Il diffère des autres corpora existants par le fait que les sujets enregistrés sont des patients souffrant de SDE et que leur niveau de somnolence est mesuré de manière subjective mais aussi objective. Le système proposé permet détecter la somnolence objective grâce à des paramètres vocaux simples et explicables à des non spécialistes.

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 nstantaneous 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 omposed 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.

La détection automatique de la somnolence peut aider le suivi de patients souffrant de maladies neuro-psychiatriques chroniques. Des recherches précédentes ont déjà montré que cela est possible en utilisant des enregistrements vocaux. Dans cet article, nous proposons d’étudier les erreurs de lecture effectuées par des patients souffrant de Somnolence Diurne Excessive (SDE) sur le corpus TILE, enregistré à l’hôpital de Bordeaux. Avec des orthophonistes, nous avons défini et compté les erreurs de lecture des patients et les avons confrontées aux différentes mesures de somnolence du corpus. Nous montrons ici que relever ces erreurs peut être utile pour élaborer des marqueurs robustes de la somnolence objective mais aussi pour définir des critères d’exclusion des locuteurs n’ayant pas un niveau de lecture suffisant.