公开(公告)号：EP0333425A2

公开(公告)日：1989-09-20

申请号：EP89302481.0

申请日：1989-03-14

申请人： UNIVERSITY OF SURREY

发明人： Evans, Barry G. Dept. of Electronic & Elect. Eng. ,  Kondoz, Ahmet M. Dept. of Electronic & Elect. Eng.

IPC分类号： G10L9/14

CPC分类号： G10L19/06 ,  G10L19/04

摘要： Speech for transmission over a low bit rate channel of a telecommunications link is digitally sampled and subjected to linear predictive coding in a coder (6). The LPC residual signal is passed through a weighting filter (18) and the output down sampled by decimation factor d resulting in d sequences of which the maximum energy sequence is selected for vector quantization. The data output from the vector quantizer (22) is formed into an output frame together with an index ( s ) identifying the selected decimation sequence and the quantized LPC parameters. The frame transmitted over the telecommunications link (46) is decoded by means of an inverse vector quantizer (50) which restores the decimated sequence. An interpolator (52) interpolates zeros to restore an excitation signal for an LPC synthesis filter (54) from which the voice signal can be reconstructed.

摘要翻译： 通过电信链路的低比特率信道进行传输的语音被数字采样并在编码器（6）中进行线性预测编码。 LPC残余信号通过加权滤波器（18），并且通过抽取因子d向下采样的输出导致其中选择了用于矢量量化的最大能量序列的d个序列。 从向量量化器（22）输出的数据与标识所选抽取序列和量化的LPC参数的索引一起形成输出帧。 通过电信链路（46）发送的帧通过恢复抽取序列的逆向量量化器（50）被解码。 内插器（52）内插0以恢复用于LPC合成滤波器（54）的激励信号，从该合成滤波器（54）可以重建语音信号。

公开(公告)号：EP0333425A3

公开(公告)日：1990-02-07

申请号：EP89302481.0

申请日：1989-03-14

申请人： UNIVERSITY OF SURREY

发明人： Evans, Barry G. Dept. of Electronic & Elect. Eng. ,  Kondoz, Ahmet M. Dept. of Electronic & Elect. Eng.

IPC分类号： G10L9/14

CPC分类号： G10L19/06 ,  G10L19/04

摘要： Speech for transmission over a low bit rate channel of a telecommunications link is digitally sampled and subjected to linear predictive coding in a coder (6). The LPC residual signal is passed through a weighting filter (18) and the output down sampled by decimation factor d resulting in d sequences of which the maximum energy sequence is selected for vector quantization. The data output from the vector quantizer (22) is formed into an output frame together with an index ( s ) identifying the selected decimation sequence and the quantized LPC parameters. The frame transmitted over the telecommunications link (46) is decoded by means of an inverse vector quantizer (50) which restores the decimated sequence. An interpolator (52) interpolates zeros to restore an excitation signal for an LPC synthesis filter (54) from which the voice signal can be reconstructed.