| 摘要 |
PROBLEM TO BE SOLVED: To generate a frame voice decision with respect to the incoming voice signal having periods of an activated voice and an unactivated voice by performing the frame voice decision of the incoming voice signal every frame according to parameters. SOLUTION: The input to a voice encoder 110 is a digitized incoming voice signal 105. Then, parameters of preliminarily determined pairs are extracted every frame from the incoming voice signal and the frame voice decision of the input voice signal is performed every frame according to the pair having different measured values extracted from the preliminarily determined pairs. The parameters of preliminarily determined pairs include the description of the spectrum of the incoming voice signal based on a line spectral frequency(LSF). Other parameters include an all band energy, a low band energy and a zero cross rate. Then, the method performing the frame voice decision of the incoming voice signal is performed by finding the sum set of subspaces are moreover each subspace is described by a linear function having at least one pair of parameters. |
