fir_test.py

#!/usr/bin/env python2

import numpy
import scipy.signal
import matplotlib.pyplot as plt

f_samp = 100.0
f_sine = 40.0

f_lim = 2.0
trans_width = 2.0



#data = numpy.sin(2 * numpy.pi * f_sine * numpy.arange(0, 300.0, 1/f_samp))
data = numpy.random.uniform(size=30000)

data_freqs = numpy.fft.rfftfreq(data.size, d=1/f_samp)
data_fft = numpy.fft.rfft(data - numpy.mean(data), norm='ortho')

coeffs = scipy.signal.firls(63, (0, f_lim, f_lim + trans_width, 50.0), (1.0, 1.0, 0.0, 0.0), nyq=f_samp/2)
filtered = scipy.signal.lfilter(coeffs, [1.0], data)


filtered_freqs = numpy.fft.rfftfreq(filtered.size, d=1/f_samp)
filtered_fft = numpy.fft.rfft(filtered - numpy.mean(filtered), norm='ortho')

ax_freq = plt.subplot(3,1,1)
ax_fft = plt.subplot(3,1,2)
ax_data = plt.subplot(3,1,3)


ax_fft.plot(data_freqs, data_fft)
ax_fft.plot(filtered_freqs, filtered_fft)

ax_data.plot(data)
ax_data.plot(filtered)

freq, response = scipy.signal.freqz(coeffs)
ax_freq.semilogy(50.0*freq/(numpy.pi), numpy.abs(response))
plt.show()