Acoustic sleepiness detection: Framework and validation of a speech-adapted pattern recognition approach

The Prediction of Fatigue Using Speech as a Biosignal

“…The optimal proposed baseline model was able to achieve an unweighted accuracy (UA) of 70.3%, and the winner of the challenge achieved an UA of 71.7% [9]. A higher UA of 82.8% has also been reported in another study based on a subset of the SLC [10].…”

“…As in [10], the relative data sparsity makes a speaker-dependent multiple hold-out cross validation approach most appropriate. Specifically a 'leave one sample out' cross validation procedure was implemented, where in each iteration a model was trained on data from all subjects with a single sample withheld for validation.…”

Measures to counteract the negative effects of night work

“…Ongoing research within this field may result in more and better real-time devices that detect sleepiness in a reliable, non-invasive manner, which can be applied in a wide area of settings (154). In addition to these oculomotoric measures, sleepinessdetection devices based upon steering analysis/lane departure warning systems (155) and speech analysis (156) have also showed promising results.…”

Sleepiness detection from speech by perceptual features

“…Using speech for sleepiness detection is one of the challenging topics in the literature, because it is a more robust configuration against environmental conditions [2,3]. Some of the related work in the literature deals with the feature extraction while others focus on classification methods to improve the detection performance.…”

“…Some of the related work in the literature deals with the feature extraction while others focus on classification methods to improve the detection performance. In [3] total of 8500 prosody, articulation and speech quality related features are calculated for detecting accident-prone fatigue state classification. The highest class wised averaged rate achieved is reported as over 80%.…”