Masanori Akita – firstname.lastname@example.org
Faculty of Engineering
700 Dannoharu Oita
Popular version of paper 1aScb9
Presented Monday morning, June 3, 2013
ICA 2013 Montreal
This paper shows the detection method of the sleep-in sleep state or sleepiness from the sound signals in the human body. In these cases, brain waves or the wave of finger plethysmogram are often used as the detection cue. However, the measurement of waves has a large load to the human body at the time of measurement and the detection methods use complicated systems and the systems using these parameters require fairly large chaotic calculations (E. Fujita et al., 2004). In a previous report we showed the detection of the sleepiness concerns with the piezoelectric sensors attached on car seat. Our preliminary examination showed that the spectrum of signals from the piezoelectric sensor have the tendency that the shapes of the spectral envelopes are flattened (S. Maeda et al., 2004 and M. Akita et al., 2005). And the sounds in the human body are considered to have similar features (M. Akita et al., 2009).
In this experiment, the signals of the piezoelectric sensor on the seat and the sounds in the human body are measured at the same time and the relationship between the sounds and sleepiness are examined. The sounds inside the body are measured using a NAM microphone system shown in Figure 1. The measurement of the signals using a piezoelectric sensor under the seat is also done. The spectral envelopes of the signals from the left side and the right side of the breath are calculated. The spectral envelopes from the seat are calculated at the same time.
Figure 1: Example photographs of the NAM microphones seat attached the piezoelectric sensor used in this experiment
In the experiment, 27 measurements by four examinees are done and 11 sleeping data are measured. The feature parameter expressing the flatness of the envelopes is defined using the lower order of cepstral coefficients and the increase of the flattened spectrum is observed by the sounds from the sleeping data of about three quarters. The detection rules are defined using the time variation of the number of the flattened spectrum. And using the rules, the detection rate of the sleepiness by the sleeping data is about 45% when the signals are obtained from the sound of the chest.
(1) E. Fujita, Y. Ogura, N. Ochiai, E. Yasuda, S. Tsuchiya, K. Murata, T. Kamei, Y. Ueno and S. Kaneko (2004). Development of simplified appraisal method of fatigue on sitting for extended periods by the data of finger plethysmogram, The Japanese Journal of Ergonomics Vol.40 No.5 254-263.
(2) S. Maeda, N. Ochiai, Y. Ogura, Y. Enoki, E. Fujita, K. Murata, T. Kamei, Y. Ueno and S. Kaneko (2004). “The simple measuring method of bio signal from buttocks”, Proceedings of the 37th Meetings Japan Ergonomics Society of Chugoku and Shikoku chapter, 8-9.
(3) M. Akita and Y. Midorikawa (2005). ” Cepstral analysis of the signals obtained from car Seat for the prediction of sleep in sleep-wake state”, Proceedings of the11th International Symposium on Applied Electromagnetics and Mechanics, 348 -349.
(4) M. Akita and Y. Midorikawa (2009). “The basic measurement of the sounds in human body for detecting the sleep-in sleep state”, Proceeding of the 10th Western Pacific Acoustic Conference, 0160 (CD ROM) 6 Pages.