发明名称 SPEECH ENCODING AND DECODING SYSTEM
摘要 PURPOSE:To reproduce an excellent speech when a circumferential noise is superposed by calculating the amplitude and position of a multipulse when the speech signal of a frame is voided and calculating and outputting only the amplitude of the pulse when the signal is voiceless. CONSTITUTION:A discrete speech signal is inputted from an input terminal 500 and a spectrum parameter, a pitch coefficient, and a pitch period are found by a spectrum, pitch, and parameter calculating circuit 520 are found and quantized by a parameter quantizer 525. A voiced/voiceless decision circuit 555 outputs voiced/voiceless information. When the speech signal of a frame is voiced, the amplitude and position of the multipulse are calculated from the cross correlation calculated from the speech signal, spectrum parameter, and pitch parameter and when the signal is voiceless, only the amplitude of the pulse is calculated at positions determined at intervals of a predetermined number of samples calculated from the cross correlation and sent out to a reception side. On a reception side, it is received to reproduce an excellent speech when the ambient noise is superposed.
申请公布号 JPH02170199(A) 申请公布日期 1990.06.29
申请号 JP19880326805 申请日期 1988.12.23
申请人 NEC CORP 发明人 OZAWA KAZUNORI
分类号 G10L19/08;G10L19/10 主分类号 G10L19/08
代理机构 代理人
主权项
地址