VLSI Design of Speaker Identification

• Main Authors: Chang, Lyang-Yu, 張良羽
• Other Authors: Tseng Hsien-Hui
• Format: Others
• Language: zh-TW
• Published: 1996
• Online Access: http://ndltd.ncl.edu.tw/handle/06762396449241643527

碩士 === 逢甲大學 === 自動控制工程研究所 === 84 === This thrsis develops an extracting algorithm to use Mel- cepstrom as speechfeature. This algorithm is used in text independent speaker identification modified from Gaussian Mixture Model and is suitable for chip design. We can get personal speech character from computer and download this character to ROMor EPROM for reference data. This chip combines with signal processing and memory to complete speaker identification system. Speaker can be identified immediately because of the high speed performance of chip to increase its practicability. This chip can identify seven people. It needs about two seconds of length of test speech. And the output is shown by 3 bits. The work speed is up to 30 MHz and the process is 0.8um CMOS technique. This chip has been send to CIC for fabrication.