def BlackHoles(windowSize=2000, step=100):
""" Task 9.5 """
file = r"sounds/BlackHolesCollision.wav"
sound, fs = Sound.Read(file)
Sound.Play(sound, fs)
sound = sound[:,
0] # The input signal has two channels (amplitude and time) - we'll need just the amplitude
N = len(sound)
X, Y, Z = [], [], [] # For the contour plot
i = 0 # Index of the window beginning
while i + windowSize < N:
window = sound[i:(i + windowSize)]
ft = np.fft.fft(window) / windowSize
#ft = FourierTransform(window)
frequencies, amplitudes = AmplitudeSpectrum(ft, fs)
X.append(np.linspace(i / fs, i / fs, windowSize // 2))
Y.append(frequencies)
Z.append(amplitudes)
i += step
plt.contourf(np.array(X), np.array(Y), np.array(Z), cmap=cm.hot)
plt.xlabel("$t$ [s]")
plt.ylabel("$f$ [Hz]")
plt.colorbar()
plt.ylim(0, 500)
plt.show()
def FTMethodComparison():
""" Task 9.6 """
file = r"sounds/a.wav"
sound, fs = Sound.Read(file)
Sound.Play(sound, fs)
sound = sound[3000:5000]
N = len(sound)
ft1 = FourierTransform(sound)
plt.plot(*AmplitudeSpectrum(ft1, fs), label="FourierTransform")
ft2 = np.fft.fft(sound) / N
plt.plot(*AmplitudeSpectrum(ft2, fs), label="numpy.fft.fft")
plt.xlabel("$f$ [Hz]")
plt.ylabel("$A$")
plt.legend()
plt.show()
def Vowels(part):
""" Task 9.4 """
path = r"sounds/"
files = ["a.wav", "e.wav", "i.wav", "o.wav", "u.wav"]
for file in files:
sound, fs = Sound.Read(path + file)
Sound.Play(sound, fs)
sound = sound[part]
ft = FourierTransform(sound)
plt.plot(*AmplitudeSpectrum(ft, fs), label=file)
plt.xlim(
0, 2000
) # Just the lowest part of the spectrum (it contains the important frequencies)
plt.xlabel("$f$ [Hz]")
plt.ylabel("$A$")
plt.legend()
plt.show()
