Sleepiness detection on read speech using simple features

Martin, Vincent P.; Rouas, Jean-Luc; Thivel, Pierre; Krajewski, Jarek

doi:10.1109/sped.2019.8906577

Cited by 13 publications

(27 citation statements)

References 22 publications

(27 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Moreover, these two modalities were detected with different vocal features, namely prosody features for spontaneous speech and formant-based features for readings. This is in line with the observation made in Martin et al ( 21 ) on the read parts of the SLC: the features have to be adapted to the type of recorded vocal production. In conclusion, both paradigms have benefits and harms: while reading texts keeps the advantage to guarantee a minimum and comparable vocal production along with the recording sessions, spontaneous speech benefits from its closeness with ecological conditions.…”

Section: Guidelinessupporting

confidence: 89%

“…The best performances have been obtained by a system based on the ASIMPLS algorithm ( 19 , 20 ) and reach 71.7%. More recently, a work focusing on the longer reading tasks of the SLC has reported an accuracy of 76.4% ( 21 ). Even if this study shows a significant improvement on a subset of the SLC, this performance is still below the necessary 80–85% for medical uses.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

How to Design a Relevant Corpus for Sleepiness Detection Through Voice?

Martin

Rouas

Micoulaud‐Franchi

et al. 2021

Front. Digit. Health

Self Cite

View full text Add to dashboard Cite

This article presents research on the detection of pathologies affecting speech through automatic analysis. Voice processing has indeed been used for evaluating several diseases such as Parkinson, Alzheimer, or depression. If some studies present results that seem sufficient for clinical applications, this is not the case for the detection of sleepiness. Even two international challenges and the recent advent of deep learning techniques have still not managed to change this situation. This article explores the hypothesis that the observed average performances of automatic processing find their cause in the design of the corpora. To this aim, we first discuss and refine the concept of sleepiness related to the ground-truth labels. Second, we present an in-depth study of four corpora, bringing to light the methodological choices that have been made and the underlying biases they may have induced. Finally, in light of this information, we propose guidelines for the design of new corpora.

show abstract

Section: Guidelinessupporting

confidence: 89%

Section: Introductionmentioning

confidence: 99%

How to Design a Relevant Corpus for Sleepiness Detection Through Voice?

Martin

Rouas

Micoulaud‐Franchi

et al. 2021

Front. Digit. Health

Self Cite

View full text Add to dashboard Cite

show abstract

“…Moreover, inspired by a previous study [18], we represented in Figure 2c) the performances of the system (a) depending on the threshold to distinguish SL from NSL. This graph represents the specificity of the selected features to the phenomena they aim at measuring.…”

Section: Performances Of the Selected Systemmentioning

confidence: 99%

“…The acoustic features used in this study are presented in detail in [17,18]. They are twofold: on one side, statistics (length and ratio) of voiced and vocalic parts are automatically extracted from audio; on the other side acoustic features are computed on these parts (Harmonics and Formants amplitude and bandwidth, Harmonic to Noise ratio, ...).…”

Section: Acoustic Featuresmentioning

confidence: 99%

Automatic Speech Recognition Systems Errors for Objective Sleepiness Detection Through Voice

Martin¹,

Rouas²,

Boyer³

et al. 2021

Interspeech 2021

Self Cite

View full text Add to dashboard Cite

Chronic sleepiness, and specifically Excessive Daytime Sleepiness (EDS), impacts everyday life and increases the risks of accidents. Compared with traditional measures (EEG), the detection of objective EDS through voice benefits from its ease to be implemented in ecological conditions and to be sober in terms of data processing and costs. Contrary to previous works focusing on short-term sleepiness estimation, this study focuses on long-term sleepiness detection through voice. Using the Multiple Sleep Latency Test corpus, this study introduces new features based on Automatic Speech Recognition systems errors, in an attempt to replace hand-labeled reading mistakes features. We also introduce a selection feature pipeline inspired by clinical validation practices allowing ASR features to perform on par with the state-of-the-art systems on short-term sleepiness detection through voice (73.2% of UAR). Moreover, we give insights on the decision process during classification and the specificity of the system regarding the threshold delimiting the two sleepiness classes, Sleepy and Non-Sleepy.

show abstract

“…Even if previous studies have already shown that it is possible to measure subjective sleepiness through voice using extracted vocal features [2,3,4], most of them were based on the Sleepy Large Corpus introduced during the Interspeech 2011 challenge [3] which included only healthy subjects. More recently, the SLEEP corpus has been elaborated for the Interspeech 2019 challenge, paving the way to the use of deep learning in sleepiness detection through voice.…”

Section: Introductionmentioning

confidence: 99%

Using reading mistakes as features for sleepiness detection in speech

Martin¹,

Chapouthier²,

Rieant³

et al. 2020

Speech Prosody 2020

Self Cite

View full text Add to dashboard Cite

Automatic detection of sleepiness can help to improve the follow-up of patients suffering from chronic diseases. Previous research on sleepiness detection has shown that this task is feasible using voice recordings. Most studies however rely on numerous features extracted from healthy subjects recordings and machine learning, the target being the output of subjective sleepiness questionnaires. In this paper, we propose to study the reading errors made by patients suffering from Excessive Daytime Sleepiness on the MSLT database collected at the Bordeaux hospital. This database differs from the others on two key points: patients are recorded instead of healthy subjects and their sleepiness level is assessed using multiple measurements, both subjective and objective. With the help of Speech Therapists, we defined and counted reading errors and confront these numbers with sleepiness measurements. We show that evaluating these reading errors can be useful to elaborate robust markers of objective sleepiness but also to elaborate exclusion criteria of the speakers not having a sufficient reading level.

show abstract

Sleepiness detection on read speech using simple features

Abstract: the queuing time and often results in episodic followups with unevenly spaced interviews.

Cited by 13 publications

References 22 publications

How to Design a Relevant Corpus for Sleepiness Detection Through Voice?

How to Design a Relevant Corpus for Sleepiness Detection Through Voice?

Automatic Speech Recognition Systems Errors for Objective Sleepiness Detection Through Voice

Using reading mistakes as features for sleepiness detection in speech

Contact Info

Product

Resources

About