Use python Write a test microphone array transfer function demo, Take what you need . This code uses a third-party library ThinkDSP.
First, let's explain what a transfer function is ：
The relationship between input and output of objects with linear characteristics , Use a function （ Ratio of Laplace transform of output waveform to Laplace transform of input waveform ） To represent the , Called transfer function .
In microphone array test , The input and output of the transfer function usually refer to the standard test signal and the signal received by each microphone . In the audio signal , The transfer function usually refers to the frequency domain , Therefore, it is necessary to perform Fourier transform on the time-domain signal （FFT）.
stay scipy There is a function class in the library signal, There's a function TransferFunction, This function can be used to calculate the transfer function between two variables .
Please refer to the following official explanations for specific usage ：
https://docs.scipy.org/doc/scipy/reference/generated/scipy.signal.TransferFunction.html
Required libraries ：
import thinkdsp
import wave
import numpy as np
import matplotlib.pyplot as plt
from scipy import signal
Ideas ：
Generate two chirp The signal s1 and s2, hold s1 As an input signal , hold s2 As if from some mic Obtained output signal . The two signals differ only in amplitude ,s2 The amplitude of is s1 Of 2 times . So according to the signal , Transfer function = Output / Input , The transfer function value of each frequency band should be 2.
First step , Generate two signals . Signal frequency range 100Hz～8kHz, Sampling rate 16kHz. The matrix elements after Fourier transform are in complex form , The transfer function calculation needs to take the real part .
s1 = thinkdsp.ExpoChirp(100,8000,1) # The amplitude is 1
wav1 = s1.make_wave(1,0,16000)
w1 = wav1.make_spectrum()
fs = w1.fs
hs1 = w1.amps
s2 = thinkdsp.ExpoChirp(100,8000,2) # The amplitude is 2
wav2 = s2.make_wave(1,0,16000)
w2 = wav2.make_spectrum()
hs2 = w2.amps
The second step , Calculate the transfer function .
s12 = signal.TransferFunction(hs2,hs1)
plt.figure('TransferFunction')
plt.plot(fs,s12.num/s12.den)
plt.ylim(0,3)
plt.show()
print('finish')
Generate the image , Of the entire frequency range 「 Output / Input 」 The transfer function values are 2, Meet preset .
Be careful , Transfer functions are not simple Output / Input , The signal processing has been done in the function TransferFunction Finish in , You can do your own research in the reference documents .
