It is necessary to provide the individual services with elderly voice in health medical tools for the elderly who aren?t accustomed to manage IT equipment. This study focuses on acoustic analysis by means of moving window method in an entire portion of the given elderly signal. Acoustic parameters based on fundamental frequency such as Jitter (%), Shimmer (%), and Signal-to-noise ratio (SNR) are calculated in each window frame using the TF32 software. The distributions estimated from female and male elderly signals overall exists abnormal ranges. Also variation between frames tends to be greater than that of normal voices. It can guess that the acoustic results vary with depending on how to set the analysis region. Therefore the decision method by the minimum perturbation with moving window method for elderly signals is recommended in this research. Moving window analysis only detects the areas where ?Err? value dropped below 10 or 20. If the acoustical parameters are estimated in the same frame, the time point is decided as the analysis section. If they have a different minimum perturbation location, then various situations must be considered for the selection of analysis section. Future investigations will develop analysis methods to improve the performance for elderly voices.
展开▼
