一、语言信号特征提取若要赋予信息处理机识别某些客观事物的能力,就需抽取事物的基本特征作为依据并加以分类。语音信号的特征提取大致可分为时域法和频域法两类。 (一)时域法图1是时域法的框图。从话筒来的语音信号经过音频放大后,从两个滤波通道输出。一个经低通滤波器以及语音过零率检测,将获得的信息送入计算机;另一 First, the language signal feature extraction To give information processors the ability to identify certain objective things, we need to extract the basic characteristics of things as a basis for classification. Speech signal feature extraction can be roughly divided into time domain method and frequency domain method two categories. (A) Time Domain Method Figure 1 is a block diagram of the time domain method. Voice signals from the microphone after audio amplification, the output from the two filter channels. A low-pass filter and voice zero-rate detection, the information obtained into the computer; the other