Python NumPy - FFT和逆FFT？

Question

所以我一直在使用FFT,我正在尝试从FFT文件中获取声音波形(最终修改它),然后将修改后的波形输出回文件.我已经获得了声波的FFT,然后在其上使用了逆FFT函数,但输出文件根本听不到.我没有对波形进行任何过滤 - 我只是测试获取频率数据然后将其放回文件中 - 听起来应该是相同的,但听起来差别很大.有任何想法吗？

- 编辑 -

从那以后我一直在研究这个项目,但还没有得到理想的结果.输出的声音文件是嘈杂的(两者都更大声,以及原始文件中不存在的额外噪声),并且来自一个声道的声音泄漏到另一个声道(之前是静音的).输入声音文件是立体声双声道文件,声音仅来自一个声道.这是我的代码:

 import scipy
 import wave
 import struct
 import numpy
 import pylab

 from scipy.io import wavfile

 rate, data = wavfile.read('./TriLeftChannel.wav')

 filtereddata = numpy.fft.rfft(data, axis=0)

 print (data)

 filteredwrite = numpy.fft.irfft(filtereddata, axis=0)

 print (filteredwrite)

 wavfile.write('TestFiltered.wav', rate, filteredwrite)

我不太清楚为什么这不起作用......？

编辑:我已经压缩了问题.py文件和音频文件,如果这可以帮助解决这里的问题.

Answer 1

1. 您似乎没有在此处应用任何过滤器
2. 你可能想采取ifft了的fft(滤波后),而不是输入波形.

Answer 2

难道不应该更像这样吗？

filtereddata = numpy.fft.fft(data)
# do fft stuff to filtereddata
filteredwrite = numpy.fft.ifft(filtereddata)
wavfile.write('TestFiltered.wav', rate, filteredwrite)

Answer 3

>>> import numpy as np
>>> a = np.vstack([np.ones(11), np.arange(11)])

# We have two channels along axis 0, the signals are along axis 1
>>> a
array([[  1.,   1.,   1.,   1.,   1.,   1.,   1.,   1.,   1.,   1.,   1.],
       [  0.,   1.,   2.,   3.,   4.,   5.,   6.,   7.,   8.,   9.,  10.]])
>>> np.fft.irfft(np.fft.rfft(a, axis=1), axis=1)
array([[  1.1       ,   1.1       ,   1.1       ,   1.1       ,
          1.1       ,   1.1       ,   1.1       ,   1.1       ,
          1.1       ,   1.1       ],
       [  0.55      ,   1.01836542,   2.51904294,   3.57565618,
          4.86463721,   6.05      ,   7.23536279,   8.52434382,
          9.58095706,  11.08163458]])
# irfft returns an even number along axis=1, even though a was (2, 11)

# When a is even along axis 1, we get a back after the irfft.
>>> a = np.vstack([np.ones(10), np.arange(10)])
>>> np.fft.irfft(np.fft.rfft(a, axis=1), axis=1)
array([[  1.00000000e+00,   1.00000000e+00,   1.00000000e+00,
          1.00000000e+00,   1.00000000e+00,   1.00000000e+00,
          1.00000000e+00,   1.00000000e+00,   1.00000000e+00,
          1.00000000e+00],
       [  7.10542736e-16,   1.00000000e+00,   2.00000000e+00,
          3.00000000e+00,   4.00000000e+00,   5.00000000e+00,
          6.00000000e+00,   7.00000000e+00,   8.00000000e+00,
          9.00000000e+00]])

# It seems like you signals are along axis 0, here is an example where the signals are on axis 0
>>> a = np.vstack([np.ones(10), np.arange(10)]).T
>>> a
array([[ 1.,  0.],
       [ 1.,  1.],
       [ 1.,  2.],
       [ 1.,  3.],
       [ 1.,  4.],
       [ 1.,  5.],
       [ 1.,  6.],
       [ 1.,  7.],
       [ 1.,  8.],
       [ 1.,  9.]])
>>> np.fft.irfft(np.fft.rfft(a, axis=0), axis=0)
array([[  1.00000000e+00,   7.10542736e-16],
       [  1.00000000e+00,   1.00000000e+00],
       [  1.00000000e+00,   2.00000000e+00],
       [  1.00000000e+00,   3.00000000e+00],
       [  1.00000000e+00,   4.00000000e+00],
       [  1.00000000e+00,   5.00000000e+00],
       [  1.00000000e+00,   6.00000000e+00],
       [  1.00000000e+00,   7.00000000e+00],
       [  1.00000000e+00,   8.00000000e+00],
       [  1.00000000e+00,   9.00000000e+00]])