OCommBIOSTEC 2019 Abstracts

Short Papers
Paper Nr: 1



Takeo Shibata

Abstract: “Pokemon Go” influenced much people to get many monsters all over the world. It gave them behaviour changes. As a consequence, they walked much everyday. We recruited Thirty volunteers to confirm if mobile phone apps with giving rewards for walking (figure 1) change their behaviour and physical abilities. Volunteers had been monitored their walking steps for one week by a step counter, then installed a mobile phone apps and monitored their walking activities for one month. The mobile phone apps shows many mascots on a map. If volunteers reach a mascot, they have some points. Some mascots give a request (for example, “Walk 500 steps in six hour.”). If the request is completed, the mascot gives more points. They can apply for a prize to get some foods. Average of daily walking steps increased from 4,653 to 7,991. Physical abilities also increased.

Paper Nr: 2

Study on Fundamental Frequency Extraction from Speech Signals of the Elderly Using Gender Innovation Method


JiYeoun Lee

Abstract: Changes in the vocal folds due to aging may change the pitch of the voice. Therefore, speech signals of elderly adults can be automatically distinguished from those of a younger demographic through analyses. Speech interfaces that are currently used in smart medical devices apply an optimized method based on the average speech patterns of young, middle-aged, and elderly adults. Deviations of the signals from this standard model may result in degradations in the performance of voice synthesis and recognition. Speech signals of elderly adults have generally been neglected in speech recognition and synthesis systems of most smart medical devices due to limitations of the speech interface. Gender innovation methods provide a basis for robust analysis of the speech differences between women and men, and this removes the effects of incorrect assumptions and stereotypes in speech analysis. In this paper, we report the sex of research subjects and the differences that exist within groups of women and men as gender analysis. The objective of this study was to extract the fundamental frequency (F0) of speech signals of elderly adults using a symmetric higher-order differential energy function based on gender innovation methods. Experiments were conducted by specifically setting the speech signals of elderly subjects as the target. The speech signals of 20 elderly Korean subjects (10 women and 10 men) ranging in age from 70 to 80 years were thus used. Symmetrical instantaneous frequency estimators with orders of five and four were respectively selected for the female and male voices in this study through gender innovation methods. The results from these experiments were compared to F0 extracted by various other methods using manual extraction, Wavesufer, TF32, and instantaneous frequency estimator based on signals before and after application of gender innovation methods. The F0 value obtained through instantaneous frequency estimator after gender analysis is the one most similar to the manual technique with an accuracy of 80%. We thus established the standards and reference models with respect to elderly speech to decrease gender bias. Our research will enhance speech recognition and synthesis performance of existing smart medical systems for the elderly. Future investigations are expected to incorporate multiple analysis methods for more reliable detection in automated medical diagnosis systems for the elderly. This will provide easy access to technologies that utilize speech for the elderly and people with vocal disabilities who may otherwise be excluded from rapid socialization.

Paper Nr: 3

The Effects of Loading on Upper Limb Physiological Tremor


Jun Akazawa and Ryuhei Okuno

Abstract: Objectives: The aim of the study was to investigate the effects of different load levels on upper limb physiological tremor in the frequency band of both the 0-4.9 Hz and the 8.0-12.9 Hz. Methods: Twelve healthy volunteers performed sustained isometric shoulder extensions at 0%, 20% and 40% maximum voluntary contraction (MVC). During the contractions, mechanical recordings were made with an acceleration sensor. Tremor power was calculated for the 0-4.9 and 8.0-12.9 Hz frequency band. Results: The both TP below 5 Hz and around 10 Hz, and RMS were increased as the weight load increased. The both TP below 5 Hz and around 10 Hz, and RMS were increased as the weight load increased. In 0% MVC, there was almost no effect of muscle contraction time. On the other hand, at other weight loads, the TP increased as the muscle contraction time increased. On the relationship between RMS and muscle contraction time at each weight load, there was almost no effect of muscle contraction time. At other weight loads, there was hardly any change except for the first 10 seconds even if the muscle contraction time increased. Conclusion: 1) Total power of frequency band below 5 Hz and around 10 Hz increased as weight load increased. 2) RMS increased as weight load increased. Regarding the influence of the muscle contraction time, 3) the total power in both frequency ranges tended to increase with the passage of time except 0% MVC. 4) It was confirmed that RMS hardly changed after passage of time as compared with the total power.