# improt library
import numpy as np
import wave
import pylab as pl
# download open-souce audio in
# http://www.voiptroubleshooter.com/open_speech/american.html
def plot_audio(audio_file='OSR_us_000_0010_8k.wav'):
# load audio
audio = wave.open(audio_file, 'rb')
# read format information
params = audio.getparams()
n_channels, _, sample_rate, n_frames = params[:4]
# read waveform data
str_wave_data = audio.readframes(n_frames)
audio.close()
# tranform waveform to array, then plot
time = np.arange(0, n_frames) * (1.0 / sample_rate)
wave_data = np.frombuffer(str_wave_data, dtype=np.short)
if n_channels == 1:
wave_data = np.reshape(wave_data, [-1, 1]).T
pl.subplot(1, 1, 1)
pl.plot(time, wave_data[0])
pl.xlabel('time (seconds)')
pl.ylabel('amplitude')
pl.show()
elif n_channels == 2:
wave_data = np.reshape(wave_data, [-1, 2]).T
pl.subplot(2, 1, 1)
pl.plot(time, wave_data[0])
pl.plot(2, 1, 2)
pl.plot(time, wave_data[1], c='r')
pl.xlabel('time (seconds)')
pl.ylabel('amplitude')
pl.show()
else:
raise ValueError('Only 1 or 2 channels supported!')