Genetic Algorithm Based Feature Selection Applied on Predicting Microsleep from Speech
Within this study we apply a speech emotion recognition engine on the detection of microsleep endangered sleepiness states. Current approaches in speech emotion recognition use low-level descriptors and functionals to compute brute-force feature sets. This paper describes an usually large feature set (45k) utilizing a broad pool of diverse elementary statistics and spectral descriptors. Several (un-)supervised subset selection methods including genetic algorithm based methods were employed on the feature space in an attempt to prune redundant dimensions. The resulting dimensionality reduced feature space was applied to speech samples gained from a car simulator based sleep deprivation study (N=12; 01.00–08.00 a.m.). Among the tested dimensionality reduction methods a simple correlation filter approach (130 features remaining) reached the best recognition rate (85.1%, SVM) in predicting microsleep endangered sleepiness stages.
KeywordsAcoustic Features Sleepiness Detection Feature Selection Genetic Algorithm
Unable to display preview. Download preview PDF.
- 1.Sommer D, Chen M, Golz M, Trutschel U, Mandic D (2005) Fusion of State Space and Frequency Domain Features for Improved Microsleep Detection. In W Dutch et al. (Eds.) Int Conf Artifical Neural Networks (ICANN 200), pp 753–759. Springer: BerlinGoogle Scholar
- 3.Harrison Y, Horne JA (1997) Sleep deprivation affects speech. Sleep 20:871–877Google Scholar
- 5.Krajewski J, Kröger B (2007) Using prosodic and spectral characteristics for sleepiness detection. Interspeech Proc., Antwerp, Belgium, 2003, pp 1841–1844Google Scholar
- 6.Krajewski J, Wieland R, Batliner A (in press) An acoustic framework for detecting fatigue in human-computer interaction. ICCHP Proc., Linz, Austria, 2008Google Scholar
- 7.Nwe TL, Li H, Dong M (2006) Analysis and Detection of Speech under Sleep Deprivation. Interspeech Proc., Pittsburgh, USA, 2006, pp 17–21Google Scholar
- 8.Vlasenk B, Schuller B, Wendemuth A, Rigoll G (2007) Combining Frame and Turn-Level Information for Robust Recognition of Emotions within Speech. Interspeech Proc., Antwerp, Belgium, pp 2249–2252Google Scholar
- 9.Batliner A, Steidl S, Schuller B, Seppi D, Laskowski K, Vogt T, Devillers L, Vidrascu L, Amir N, Kessous L, Aharonson V (2006) Combining Efforts for Improving Automatic Classification of Emotional User States. In Erjavec T & Gros JZ (Eds.): Language Technologies, IS-LTC 2006, Ljubljana, Slovenia, pp 240–245Google Scholar
- 11.Schuller B, Reiter S, Rigoll G (2006) Evolutionary Feature Generation in Speech Emotion Recognition. ICME 2006, Toronto, Canada, 2006, pp 5–8Google Scholar
- 13.Boersma P (2001) PRAAT, a system for doing phonetics by computer, Glot International 5:341–345Google Scholar
- 15.Webber C L, Zbilut J P (1994) Dynamical assessment of physiological systems and states using recurrence plot strategies. Journal of Applied Physiology 76: 965–973Google Scholar