def file_test (small_exponent = 9, large_exponent = 12):
"multigram"
s = K.Multigram(small_exponent, large_exponent)
size_in = s.size_in
large_size = s.size_out
length = s.size_out / 2#arbitrary
print "reading sound file"
sound = formats.read_wav('test.wav', size_in*length, size_in)
image = K.Reals(length, large_size)
print "transforming data"
for i in range(length):
s.transform_fwd(sound[i,:], image[i,:])
del sound
print "saving image"
image = util.energy_to_loudness(image + 1e-5)
formats.write_image(image, 'test.png')
def show_file (size_exponent = 10, time_exponent = 2, freq_exponent = 3):
'Shows supergram of an audio file'
s = K.Supergram(size_exponent, time_exponent, freq_exponent)
size = s.size
small_size = s.small_size
large_size = s.super_size
length = s.super_size #arbitrary
print "reading sound file"
sound = formats.read_wav('test.wav', small_size*length, small_size)
image = K.transforms.Reals(length, large_size)
print "transforming data"
for i in range(length):
s.transform_fwd(sound[i,:], image[i,:])
#del sound
print "saving image"
image = K.util.energy_to_loudness(image + 1e-5)
formats.write_image(image, 'test.png').show()
def show_file(bank_size=512, block_size=K.DEFAULT_FRAMES_PER_BUFFER):
"Shows pitchgram of an audio file"
size_in = block_size
size_out = bank_size
length = size_out * 16 / 9 # arbitrary
pitchgram = K.Pitchgram(size_in, size_out)
print "reading sound file"
sound = formats.read_wav("test.wav", size_in * length, size_in)
image = K.transforms.Reals(length, size_out)
print "transforming data"
for i in range(length):
pitchgram.transform(sound[i, :], image[i, :])
# del sound
print "saving image"
image = K.util.energy_to_loudness(image + 1e-5)
formats.write_image(image, "test.png").show()
def read_spectrogram (exponent = 10, wavname = 'test.wav'):
"""
read a sound file, say a wav file;
compute a spectrogram or reassigned spectrogram;
write image to png file
"""
width = 1 << exponent
height = width
size = height * width
print "reading sound file"
sound = formats.read_wav(wavname, size, width)
image = K.Reals(height, width/2)
print "transforming data"
s = K.Spectrogram(exponent)
for i in range(height):
s.transform_fwd(sound[i,:],image[i,:])
print "saving image"
image = formats.energy_to_loudness(image)
formats.write_image(image, 'test.png')
