Biological-sound analysis device, biological-sound analysis method, program, and storage medium

A biological sound analyzing apparatus, comprising:

• a processor, in communication with a memory;
• and
• an output device configured to display information output from the processor as image data,
• said processor configured to function at least as:
• an obtainer configured to obtain first information regarding biological sounds in a first period, said first information obtained as waveform information of a sound signal to be processed by said processor and indicating change of said biological sounds in said first period;
• a division processor configured to perform a process of dividing the first information into a plurality of frame information according to a plurality of second periods, each of said second periods being shorter than the first period;
• a determiner configured to determine, for each one of the plurality of frame information, whether or not sounds indicating abnormality of a living body and included in the biological sounds are generated;
• and
• an outputter configured to output second information to said output device, said second information indicating a ratio of a generation time in which the sounds indicating the abnormality are generated, with respect to the first period, on the basis of the first information, said second information is calculated as a ratio of the frame information in which the sounds indicating the abnormality are generated, with respect to all the frame information for which determination is performed by said determiner,
• wherein the determiner determines whether or not sounds are generated indicating the abnormality included in the biological sounds, for each one of the plurality of frame information, by performing a pre-filter process on the waveform information of the sound signal and calculating a maximum value of a local variance in each frame of the plurality of frame information according to the second period to determine a first local feature amount, and by obtaining a spectrum on the waveform information of the sound signal and performing a cepstral mean normalization process, performing a liftering process to extract an envelope component, and a liftering process to extract a micro component to determine a second local feature amount,
• the first and second local feature amounts used to determine said second information indicating the ratio of the generation time in which the sounds indicating abnormality are generated, with respect to the first period, for detecting the generation of the abnormal sounds in the biological sounds.