2

我有一个音频信号输入作为双精度数组。我想通过将其分割成 25 毫秒的汉明窗口分析帧来制作该信号的帧,重叠率为 50%。我有计算汉明窗的代码。有人可以告诉我如何将音频信号分割成帧吗?首选伪代码或 Java 代码。

4

1 回答 1

3

伪代码:

Fs = 44100;                // sample rate, e.g. 44.1 kHz
overlap = 0.5;             // overlap = 50%
nw = Fs * 0.025;           // no of samples in window
ns = nw * (1.0 - overlap); // no of samples to increment n0, n1
n0 = 0;                    // window sample no begin
n1 = n0 + nw;              // window sample no end
N = 10 * Fs;               // N = total no of samples to process, e.g. 10 seconds

do
    get nw samples from n0 to (n1 - 1) into temporary buffer
    apply window function to temporary buffer
    apply FFT to temporary buffer
    ... etc ...
    n0 += ns; // increment window start/end by no of overlap samples
    n1 += ns;
while n1 <= N;
于 2012-05-15T16:52:03.897 回答