海明窗

　语音信号一般在10ms到30ms之间，我们可以把它看成是平稳的。为了处理语音信号，我们要对语音信号进行加窗，也就是一次仅处理窗中的数据。因为实际的语音信号是很长的，我们不能也不必对非常长的数据进行一次性处理。明智的解决办法就是每次取一段数据，进行分析，然后再取下一段数据，再进行分析。 -Hamming window voice signal is generally between 10ms to 30ms, we can see it as is stationary. To handle voice signals, we add window to the speech signal, that is, once a process only the data in the window. Because the actual speech signal is very long, we can not do not have a very long data one-time processing. The sensible solution is to always take a piece of data, analysis, and then remove the piece of data, and then analyzed.