| 摘要 |
PURPOSE:To enable the usage under high surrounding noise atomosphere, by analyzing the mixing signal with training audio and surrounding noise in advance and obtaining required parameter groups. CONSTITUTION:The input unit 109 receives the training audio and external noise 108 from the audio output unit 107, and gives the output to the parameter analyzer 110. The classification memory unit 112 classifies and memorizes the parameter from the analyzer 110 based on the classified data of training audio from the data file 111. The discrimination count measurement unit 113 determines the coefficient of the linear discrimination function from the parameter classified and memorized and the threshold value of classification and stores them to the memory unit 114. The input unit 116 receives audio and noise 108 from the caller 115 and inputs the output to the analyzer 117. The analyzer 117 gives the parameter of the signal from the input unit 116 to the coder 118 based on the data of the memory unit 114. The synthesizer 103 preforms audio synthesis with the parameter by means of well known methods. |
