def preprocess_arabic(text, preprocess_config, bw = False):
text = text.rstrip(punctuation)
if bw:
text = "".join([bw2ar[l] if l in bw2ar else l for l in text])
phones = ''
for word in text.split(' '):
if word in punctuation:
pass
elif len(word.strip()) > 0:
phones+=phonetise(word)[0]
phones = "{" + "}{".join(phones.split(' ')) + "}"
phones = phones.replace("}{", " ")
print("Raw Text Sequence: {}".format(text))
print("Phoneme Sequence: {}".format(phones))
sequence = np.array(
#TO_DO
text_to_sequence(
phones, preprocess_config["preprocessing"]["text"]["text_cleaners"]
)
)
return np.array(sequence)
def _maybe_get_arpabet(word):
pronunciations = phonetise(word)
toBeReturned = '{%s}' % pronunciations[0] if len(
pronunciations) == 1 else '{%s}' % pronunciations[1]
return toBeReturned
def test_2():
actual = phonetise('ثٌمّ')
expected = ['^ u0 n mm']
assert actual == expected
def test_1():
actual = phonetise('ثٌمَّ')
expected = ['^ u0 n mm a']
assert actual == expected
def test_3():
actual = phonetise('ثكمّ')
expected = ['^ k mm']
assert actual == expected