def test_find_stress(self):
test_string = "reflect respect recline reduce obsessively demonstrate baseball cloud brother cobblestone " +\
"complete conspire estuary"
raw_cmu = transcribe.get_cmu(test_string.split(" "))
result = []
for word_list in raw_cmu:
for word in word_list:
stressed = stress.find_stress(word)
result.append(stressed)
self.assertEqual(result, [
'r ah ˈf l eh k t', 'r ih ˈf l eh k t', 'r ih ˈs p eh k t',
'r iy ˈs p eh k t', 'r ih ˈk l ay n', 'r ih ˈd uw s',
'aa b ˈs eh s ih v l iy', 'ˈd eh m ah n ˌs t r ey t',
'ˈb ey s ˈb ao l', 'k l aw d', 'ˈb r ah dh er',
'ˈk aa b ah l ˌs t ow n', 'k ah m ˈp l iy t', 'k ah n ˈs p ay er',
'ˈeh s ch uw ˌeh r iy'
])
# test the retrieval of only primary stress
self.assertEqual(
stress.find_stress("d eh1 m ah0 n s t r ey2 t", type="primary"),
'ˈd eh m ah n s t r ey t')
# test the retrieval of only secondary stress
self.assertEqual(
stress.find_stress("d eh1 m ah0 n s t r ey2 t", type="secondary"),
'd eh m ah n ˌs t r ey t')
def cmu_to_ipa(cmu_list, mark=True, stress_marking='all'):
"""converts the CMU word lists into IPA transcriptions"""
symbols = {"a": "ə", "ey": "e", "aa": "ɑ", "ae": "æ", "ah": "ə", "ao": "ɔ",
"aw": "aʊ", "ay": "aɪ", "ch": "ʧ", "dh": "ð", "eh": "ɛ", "er": "ər",
"hh": "h", "ih": "ɪ", "jh": "ʤ", "ng": "ŋ", "ow": "oʊ", "oy": "ɔɪ",
"sh": "ʃ", "th": "θ", "uh": "ʊ", "uw": "u", "zh": "ʒ", "iy": "i", "y": "j"}
ipa_list = [] # the final list of IPA tokens to be returned
for word_list in cmu_list:
ipa_word_list = [] # the word list for each word
for word in word_list:
if stress_marking:
word = stress.find_stress(word, type=stress_marking)
else:
if re.sub("\d*", "", word.replace("__IGNORE__", "")) == "":
pass # do not delete token if it's all numbers
else:
word = re.sub("[0-9]", "", word)
ipa_form = ''
if word.startswith("__IGNORE__"):
ipa_form = word.replace("__IGNORE__", "")
# mark words we couldn't transliterate with an asterisk:
if mark:
if not re.sub("\d*", "", ipa_form) == "":
ipa_form += "*"
else:
for piece in word.split(" "):
marked = False
unmarked = piece
if piece[0] in ["ˈ", "ˌ"]:
marked = True
mark = piece[0]
unmarked = piece[1:]
if unmarked in symbols:
if marked:
ipa_form += mark + symbols[unmarked]
else:
ipa_form += symbols[unmarked]
else:
ipa_form += piece
swap_list = [["ˈər", "əˈr"], ["ˈie", "iˈe"]]
for sym in swap_list:
if not ipa_form.startswith(sym[0]):
ipa_form = ipa_form.replace(sym[0], sym[1])
ipa_word_list.append(ipa_form)
ipa_list.append(sorted(list(set(ipa_word_list))))
return ipa_list