def test_minmax_array_numpy(self):
test_sound_1 = SoundPreprocessor(TestSoundPreprocessor.voicefile_1)
result = test_sound_1.minmax_array_numpy()
self.assertIsInstance(result, ndarray)
self.assertIsInstance(result.shape, tuple)
self.assertEqual(len(result.shape), 1)
self.assertAlmostEqual(result.min(), 0.0)
self.assertAlmostEqual(result.max(), 10.0)
def test_create_voice_image_mean_array(self):
test_sound_1 = TestSoundPreprocessor.create_test_sound(
TestSoundPreprocessor.voicefile_1)
test_sound_2 = TestSoundPreprocessor.create_test_sound(
TestSoundPreprocessor.voicefile_2)
result = SoundPreprocessor.create_voice_image_mean_array(
[test_sound_1.scipy_audio, test_sound_2.scipy_audio])
self.assertIsInstance(result, ndarray)
self.assertIsInstance(result.shape, tuple)
self.assertEqual(len(result.shape), 1)
self.assertAlmostEqual(result.min(), 0.0)
self.assertAlmostEqual(result.max(), 10.0)
def post(self, merchant_id: int, user_id: int, text_id: int, local_filename: str, remote_filename: str):
"""
create an ndarray out of .wav file sample and upload it to database
:return: json message
"""
# initialize voice array model from input data
new_voice_array = VoiceArrayModel(merchant_id, user_id, text_id, local_filename)
new_voice_array_exist = os.path.isfile(os.path.join(WorkingFolders.upload_folder, local_filename))
if not new_voice_array_exist:
return {'message': f'Filename: {local_filename} does not exists on back-end server!',
'status': 'error'}, 404
try:
# transform input wavefile
input_sound = SoundPreprocessor(os.path.join(WorkingFolders.upload_folder, local_filename))
# input_sound.convert_stereo_to_mono()
input_sound.fourier_transform_audio()
input_sound.minmax_array_numpy()
except Exception as e:
return e
# save .npy file
save(os.path.join(WorkingFolders.arrays_folder, remote_filename), input_sound.scipy_audio)
# upload to database as binary
result = UploadFileToDatabase.post(os.path.join(WorkingFolders.arrays_folder, remote_filename))
# delete local array
os.remove(os.path.join(WorkingFolders.arrays_folder, remote_filename))
os.remove(os.path.join(WorkingFolders.upload_folder, new_voice_array.local_filename))
return {'message': result}
def generate_binary_voice_image(arrays_list: list, local_filename: str):
"""
generates binary image from average values of voice arrays (per specific text) and upload it up to database
:return: bool
"""
# create ndarray from selected arrays
image_ndarray = SoundPreprocessor.create_voice_image_mean_array(
arrays_list)
# generate image out of compiled ndarray file
_, stored_image_buffer = ImagePreprocessor.generate_audio_image(
image_ndarray)
with open(local_filename, "wb") as f:
f.write(stored_image_buffer.getbuffer())
return local_filename
def verify_voice(user_email: str, local_wavefile: str,
local_voice_image: str):
# process input sound
input_sound = SoundPreprocessor(local_wavefile)
# input_sound.convert_stereo_to_mono()
input_sound.fourier_transform_audio()
input_sound.minmax_array_numpy()
# generate image from processed audio and put it into buffer
input_image_buffer: BytesIO
_, input_image_buffer = ImagePreprocessor.generate_audio_image(
input_sound.scipy_audio)
# compare images
image_preprocessor = ImagePreprocessor(input_image_buffer,
local_voice_image)
result_dhash = image_preprocessor.compare_dhash()
result_whash = image_preprocessor.compare_whash()
print(f"DHASH Difference: {result_dhash}")
print(f"WHASH Difference: {result_whash}")
# close BytesIO buffers
input_image_buffer.close()
if result_dhash > 600 or result_whash > 600:
return {
'message': 'Values over 600!',
'dhash': str(result_dhash),
'whash': str(result_whash),
'status': 'error'
}, 403
elif result_dhash > result_whash:
return {
'message': 'DHASH bigger then WHASH!!',
'dhash': str(result_dhash),
'whash': str(result_whash),
'status': 'error'
}, 403
elif result_whash / result_dhash < 1.5:
return {
'message': 'Proportion lower than 1.5!!',
'dhash': str(result_dhash),
'whash': str(result_whash),
'status': 'error'
}, 403
else:
if result_dhash < 250:
if result_dhash <= 220:
return {
'message': f'Minimal DHASH OK: {result_dhash}',
'dhash': str(result_dhash),
'whash': str(result_whash),
'status': 'success'
}, 200
elif result_dhash + result_whash < 650:
return {
'message': f'Sum OK: {result_dhash + result_whash}',
'dhash': str(result_dhash),
'whash': str(result_whash),
'status': 'success'
}, 200
else:
return {
'message':
f'Sum too big: {result_dhash + result_whash}',
'dhash': str(result_dhash),
'whash': str(result_whash),
'status': 'error'
}, 403
else:
return {
'message': f'DHASH too big: {result_dhash}',
'dhash': str(result_dhash),
'whash': str(result_whash),
'status': 'error'
}, 403
def create_test_sound(imagefile):
test_sound = SoundPreprocessor(imagefile)
test_sound.fourier_transform_audio()
test_sound.minmax_array_numpy()
return test_sound
def test_fourier_transform_audio(self):
test_sound_2 = SoundPreprocessor(TestSoundPreprocessor.voicefile_2)
result = test_sound_2.fourier_transform_audio()
self.assertIsInstance(result, ndarray)
self.assertIsInstance(result.shape, tuple)
self.assertEqual(len(result.shape), 1)
def test_return_bit_depth(self):
test_sound_1 = SoundPreprocessor(TestSoundPreprocessor.voicefile_1)
result = test_sound_1.return_bit_depth()
self.assertIsInstance(result, int)
def test_voice_2(self):
test_sound_2 = SoundPreprocessor(TestSoundPreprocessor.voicefile_2)
self.assertIsInstance(test_sound_2, object)