PURPOSE: To provide an acoustic detection device capable of presenting a description of a dynamically changing signal from a sound source with a simple code line and reducing a data amount and an operation amount required for pattern matching.
CONSTITUTION: A sound print forming part 101 forms a sound crest display from an inputted acoustic signal PS, and after a binarization processing part 102 makes it an image, a processing part for thinning line 103 extracts the central line of the image, and an encoding processing part 104 obtains a characteristic of a frequency shift of each line at the point of time while referring to a decision rule of a line characteristic of a decision rule part 110 of the line characteristic. A listing processing part 105 forms a collating list (code line) one-dimensionally, and a collation processing part 106 extracts the characteristics (shape, observation time) by collating the decision rule (knowledge base) of the signal characteristic the decision rule part 120 of the signal characteristic, and a target identification part 107 gains the characteristic of the acoustic signal PS, and it identifies a target sound source.
JPS61120185 | MUSICAL SOUND ANALYZER |
JP3687135 | SOUND OSCILLATION EVALUATING DEVICE |
AMAMIYA HIDEKI
YAMAGUCHI ISAO
TAKAGI KENJI
TSUJIMARU ATSUMI
NEC CORP
NEC AEROSPACE SYS LTD