在做一个scipy.signal.spectrogram的时候，怎么知道会有多少个时间段呢？

Question

我有：

    frequencies, times, spectrogram = signal.spectrogram(
        samples, sample_rate, nperseg=nperseg, window=signal.hann(nperseg), noverlap=noverlap, mode='magnitude')

我有 nperseg=320 和 noverlap=80，还有一个 sample_rate=16000。

我有 232800 个样本。

但是，我的 spectrogram.shape 现在是 (161, 969)。所以这是 161 个频率仓和 969 个时间段。那969是怎么算出来的？

Answer 1

    time_per_segment_ms = 20
    nperseg = int(sample_rate * 0.001 * time_per_segment_ms)
    overlap = nperseg // 4

    seconds_per_segment = (nperseg - overlap) / sample_rate
    ms_per_segment = int(seconds_per_segment * 1000)

差不多就可以了

在做一个scipy.signal.spectrogram的时候，怎么知道会有多少个时间段呢？

When doing a scipy.signal.spectrogram, how can I tell how many time segments there will be?

python

fft

scipy

spectrogram