作者: Masayuki Nishiguchi
DOI: 10.1121/1.423205
关键词: Computer science 、 Speech recognition 、 Background noise 、 Value (computer science) 、 Block (data storage) 、 Acoustics 、 Frame (networking) 、 Standard deviation
摘要: A method and a device for discriminating voiced sound from an unvoiced or background noise in speech signals are disclosed. Each block frame of input is divided into plural sub-blocks the standard deviation, effective value peak detected detection unit detecting statistical characteristics one sub-block to another. bias detects on time scale decide whether