使用定义的 FFT 长度、块大小和波形文件的窗口计算 FFT

问题描述

我有一个以 48KHz 采样率记录 5 秒的 32 位浮点 .wav 文件。我想在应用汉宁窗的情况下获得完整 5 秒的 FFT,没有重叠,并且 FFT 长度为 8192。在获得 FFT 时,我意识到我在 FFT 计算中犯了一个错误。

我使用以下代码来执行此操作:

# Imports
import matplotlib.pyplot as plt
import numpy as np
import math as math
import soundfile as sf  
import scipy.fftpack as fftpack
import scipy.io.wavfile as wf


def dbfft(x,fs,win=None,ref=1):
    """
    Calculate spectrum in dB scale
    Args:
        x: input signal
        fs: sampling frequency
        win: vector containing window samples (same length as x).
             If not provided,then rectangular window is used by default.
        ref: reference value used for dBFS . 32768 for int16 and 1 for float

    Returns:
        freq: frequency vector
        s_db: spectrum in dB scale
    """
    N = len(x)  # Length of input sequence
    if win is None:
        win = np.ones(1,N)
    if len(x) != len(win):
            raise ValueError('Signal and window must be of the same length')
    x = x * win
    # Calculate real FFT and frequency vector
    sp = np.fft.rfft(x)
    freq = np.arange((N / 2) + 1) / (float(N) / fs)
    # Scale the magnitude of FFT by window and factor of 2,because we are using half of FFT spectrum.
    s_mag = np.abs(sp) * 2 / np.sum(win)
    # Convert to dBFS using 20*log10(val/max)
    s_dbfs = 20 * np.log10(s_mag/ref)      
    return freq,s_dbfs

def main():
    # Read from wav file
    data,samplerate = sf.read('Signal Analyzer_5s_55_1_L1F4_APxRecordedAudio.wav')
    # Scaling factor
    K = 120
    # Calculation
    N = 8192
    win = np.hanning(N)
    # Frequency domain
    freq,s_dbfs = dbfft(data[0:N],samplerate,win)
    # Scale from dbFS to dB
    s_db = s_dbfs + K
    # Time domain
    Time = np.linspace(0,len(data) / samplerate,num=len(data))
    # Amp_time = 20 * np.log10 (abs(data) / 0.00002)    # ref = 1 
    Amp_time = (20*np.log10((data/50))) + K + 20*np.log10(math.sqrt(2))      # reference of sound i.e 20 log10(P rms_value/P ref)+120 dB TODO
  

    # Plot
    #fig,axes = plt.subplots(nrows=2,ncols=1)
    plt.subplot(2,1,1)
    plt.plot(Time,Amp_time)
    plt.grid('on')
    plt.minorticks_on
    plt.xlabel('Time [s]')
    plt.ylabel('Instantaneous Level [dBFS]')
    plt.xlim([0,5])
    plt.subplot(2,2)
    plt.plot(freq,s_db)
    plt.grid('on')
    plt.minorticks_on
    plt.xlabel('Frequency [Hz]')
    plt.ylabel('Amplitude [dB]')
    plt.xscale('log')
    plt.xlim([10,10000])
    plt.show()
    
if __name__ == "__main__":
    main()

在代码中,我看到我只对前 8192 个样本进行 FFT,并对完整的 240000(5 秒)样本进行平均 FFT,块大小为 8192,带有汉宁窗口。对于 5 秒(88 次 FFT),我是否应该每 8192 秒进行多次 FFT 并平均幅度以获得结果 FFT?有没有一种有效的方法来执行此操作?

解决方法

也许您想了解 spectrogramshort-time Fourier transform

您也许可以开始使用 librosa。例如用 specshow 绘制频谱图。

相关问答

错误1:Request method ‘DELETE‘ not supported 错误还原:...
错误1:启动docker镜像时报错:Error response from daemon:...
错误1:private field ‘xxx‘ is never assigned 按Alt...
报错如下,通过源不能下载,最后警告pip需升级版本 Requirem...