voicevox / test /test_full_context_label.py
2ndelement's picture
init
f1f433f
raw
history blame
16.1 kB
from copy import deepcopy
from itertools import chain
from unittest import TestCase
from voicevox_engine.full_context_label import (
AccentPhrase,
BreathGroup,
Mora,
Phoneme,
Utterance,
)
class TestBasePhonemes(TestCase):
def setUp(self):
super().setUp()
# pyopenjtalk.extract_fullcontext("ใ“ใ‚“ใซใกใฏใ€ใƒ’ใƒ›ใงใ™ใ€‚")ใฎ็ตๆžœ
# ๅ‡บๆฅใ‚‹้™ใ‚Šใƒ†ใ‚นใƒˆๅ†…ใงไป–ใฎใƒฉใ‚คใƒ–ใƒฉใƒชใซไพๅญ˜ใ—ใชใ„ใŸใ‚ใ€
# ใพใŸใƒ†ใ‚นใƒˆๅ†…ๅฎนใ‚’้€ๆ˜ŽๅŒ–ใ™ใ‚‹ใŸใ‚ใซใ€ใƒ†ใ‚นใƒˆใ‚ฑใƒผใ‚นใ‚’็”Ÿๆˆใ—ใฆใ„ใ‚‹
self.test_case_hello_hiho = [
# sil (็„ก้Ÿณ)
"xx^xx-sil+k=o/A:xx+xx+xx/B:xx-xx_xx/C:xx_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:xx_xx#xx_xx@xx_xx|xx_xx/G:5_5%0_xx_xx/H:xx_xx/I:xx-xx"
+ "@xx+xx&xx-xx|xx+xx/J:1_5/K:2+2-9",
# k
"xx^sil-k+o=N/A:-4+1+5/B:xx-xx_xx/C:09_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:5_5#0_xx@1_1|1_5/G:4_1%0_xx_0/H:xx_xx/I:1-5"
+ "@1+2&1-2|1+9/J:1_4/K:2+2-9",
# o
"sil^k-o+N=n/A:-4+1+5/B:xx-xx_xx/C:09_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:5_5#0_xx@1_1|1_5/G:4_1%0_xx_0/H:xx_xx/I:1-5"
+ "@1+2&1-2|1+9/J:1_4/K:2+2-9",
# N (ใ‚“)
"k^o-N+n=i/A:-3+2+4/B:xx-xx_xx/C:09_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:5_5#0_xx@1_1|1_5/G:4_1%0_xx_0/H:xx_xx/I:1-5"
+ "@1+2&1-2|1+9/J:1_4/K:2+2-9",
# n
"o^N-n+i=ch/A:-2+3+3/B:xx-xx_xx/C:09_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:5_5#0_xx@1_1|1_5/G:4_1%0_xx_0/H:xx_xx/I:1-5"
+ "@1+2&1-2|1+9/J:1_4/K:2+2-9",
# i
"N^n-i+ch=i/A:-2+3+3/B:xx-xx_xx/C:09_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:5_5#0_xx@1_1|1_5/G:4_1%0_xx_0/H:xx_xx/I:1-5"
+ "@1+2&1-2|1+9/J:1_4/K:2+2-9",
# ch
"n^i-ch+i=w/A:-1+4+2/B:xx-xx_xx/C:09_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:5_5#0_xx@1_1|1_5/G:4_1%0_xx_0/H:xx_xx/I:1-5"
+ "@1+2&1-2|1+9/J:1_4/K:2+2-9",
# i
"i^ch-i+w=a/A:-1+4+2/B:xx-xx_xx/C:09_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:5_5#0_xx@1_1|1_5/G:4_1%0_xx_0/H:xx_xx/I:1-5"
+ "@1+2&1-2|1+9/J:1_4/K:2+2-9",
# w
"ch^i-w+a=pau/A:0+5+1/B:xx-xx_xx/C:09_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:5_5#0_xx@1_1|1_5/G:4_1%0_xx_0/H:xx_xx/I:1-5"
+ "@1+2&1-2|1+9/J:1_4/K:2+2-9",
# a
"i^w-a+pau=h/A:0+5+1/B:xx-xx_xx/C:09_xx+xx/D:09+xx_xx/E:xx_xx!xx_xx-xx"
+ "/F:5_5#0_xx@1_1|1_5/G:4_1%0_xx_0/H:xx_xx/I:1-5"
+ "@1+2&1-2|1+9/J:1_4/K:2+2-9",
# pau (่ชญ็‚น)
"w^a-pau+h=i/A:xx+xx+xx/B:09-xx_xx/C:xx_xx+xx/D:09+xx_xx/E:5_5!0_xx-xx"
+ "/F:xx_xx#xx_xx@xx_xx|xx_xx/G:4_1%0_xx_xx/H:1_5/I:xx-xx"
+ "@xx+xx&xx-xx|xx+xx/J:1_4/K:2+2-9",
# h
"a^pau-h+i=h/A:0+1+4/B:09-xx_xx/C:09_xx+xx/D:22+xx_xx/E:5_5!0_xx-0"
+ "/F:4_1#0_xx@1_1|1_4/G:xx_xx%xx_xx_xx/H:1_5/I:1-4"
+ "@2+1&2-1|6+4/J:xx_xx/K:2+2-9",
# i
"pau^h-i+h=o/A:0+1+4/B:09-xx_xx/C:09_xx+xx/D:22+xx_xx/E:5_5!0_xx-0"
+ "/F:4_1#0_xx@1_1|1_4/G:xx_xx%xx_xx_xx/H:1_5/I:1-4"
+ "@2+1&2-1|6+4/J:xx_xx/K:2+2-9",
# h
"h^i-h+o=d/A:1+2+3/B:09-xx_xx/C:22_xx+xx/D:10+7_2/E:5_5!0_xx-0"
+ "/F:4_1#0_xx@1_1|1_4/G:xx_xx%xx_xx_xx/H:1_5/I:1-4"
+ "@2+1&2-1|6+4/J:xx_xx/K:2+2-9",
# o
"i^h-o+d=e/A:1+2+3/B:09-xx_xx/C:22_xx+xx/D:10+7_2/E:5_5!0_xx-0"
+ "/F:4_1#0_xx@1_1|1_4/G:xx_xx%xx_xx_xx/H:1_5/I:1-4"
+ "@2+1&2-1|6+4/J:xx_xx/K:2+2-9",
# d
"h^o-d+e=s/A:2+3+2/B:22-xx_xx/C:10_7+2/D:xx+xx_xx/E:5_5!0_xx-0"
+ "/F:4_1#0_xx@1_1|1_4/G:xx_xx%xx_xx_xx/H:1_5/I:1-4"
+ "@2+1&2-1|6+4/J:xx_xx/K:2+2-9",
# e
"o^d-e+s=U/A:2+3+2/B:22-xx_xx/C:10_7+2/D:xx+xx_xx/E:5_5!0_xx-0"
+ "/F:4_1#0_xx@1_1|1_4/G:xx_xx%xx_xx_xx/H:1_5/I:1-4"
+ "@2+1&2-1|6+4/J:xx_xx/K:2+2-9",
# s
"d^e-s+U=sil/A:3+4+1/B:22-xx_xx/C:10_7+2/D:xx+xx_xx/E:5_5!0_xx-0"
+ "/F:4_1#0_xx@1_1|1_4/G:xx_xx%xx_xx_xx/H:1_5/I:1-4"
+ "@2+1&2-1|6+4/J:xx_xx/K:2+2-9",
# U (็„กๅฃฐๆฏ้Ÿณ)
"e^s-U+sil=xx/A:3+4+1/B:22-xx_xx/C:10_7+2/D:xx+xx_xx/E:5_5!0_xx-0"
+ "/F:4_1#0_xx@1_1|1_4/G:xx_xx%xx_xx_xx/H:1_5/I:1-4"
+ "@2+1&2-1|6+4/J:xx_xx/K:2+2-9",
# sil (็„ก้Ÿณ)
"s^U-sil+xx=xx/A:xx+xx+xx/B:10-7_2/C:xx_xx+xx/D:xx+xx_xx/E:4_1!0_xx-xx"
+ "/F:xx_xx#xx_xx@xx_xx|xx_xx/G:xx_xx%xx_xx_xx/H:1_4/I:xx-xx"
+ "@xx+xx&xx-xx|xx+xx/J:xx_xx/K:2+2-9",
]
self.phonemes_hello_hiho = [
Phoneme.from_label(label) for label in self.test_case_hello_hiho
]
class TestPhoneme(TestBasePhonemes):
def test_phoneme(self):
self.assertEqual(
" ".join([phoneme.phoneme for phoneme in self.phonemes_hello_hiho]),
"sil k o N n i ch i w a pau h i h o d e s U sil",
)
def test_is_pause(self):
self.assertEqual(
[phoneme.is_pause() for phoneme in self.phonemes_hello_hiho],
[
True, # sil
False, # k
False, # o
False, # N
False, # n
False, # i
False, # ch
False, # i
False, # w
False, # a
True, # pau
False, # h
False, # i
False, # h
False, # o
False, # d
False, # e
False, # s
False, # u
True, # sil
],
)
def test_label(self) -> None:
self.assertEqual(
[phoneme.label for phoneme in self.phonemes_hello_hiho],
self.test_case_hello_hiho,
)
class TestMora(TestBasePhonemes):
def setUp(self) -> None:
super().setUp()
# contexts["a2"] == "1" ko
self.mora_hello_1 = Mora(
consonant=self.phonemes_hello_hiho[1], vowel=self.phonemes_hello_hiho[2]
)
# contexts["a2"] == "2" N
self.mora_hello_2 = Mora(consonant=None, vowel=self.phonemes_hello_hiho[3])
# contexts["a2"] == "3" ni
self.mora_hello_3 = Mora(
consonant=self.phonemes_hello_hiho[4], vowel=self.phonemes_hello_hiho[5]
)
# contexts["a2"] == "4" chi
self.mora_hello_4 = Mora(
consonant=self.phonemes_hello_hiho[6], vowel=self.phonemes_hello_hiho[7]
)
# contexts["a2"] == "5" wa
self.mora_hello_5 = Mora(
consonant=self.phonemes_hello_hiho[8], vowel=self.phonemes_hello_hiho[9]
)
# contexts["a2"] == "1" hi
self.mora_hiho_1 = Mora(
consonant=self.phonemes_hello_hiho[11], vowel=self.phonemes_hello_hiho[12]
)
# contexts["a2"] == "2" ho
self.mora_hiho_2 = Mora(
consonant=self.phonemes_hello_hiho[13], vowel=self.phonemes_hello_hiho[14]
)
# contexts["a2"] == "3" de
self.mora_hiho_3 = Mora(
consonant=self.phonemes_hello_hiho[15], vowel=self.phonemes_hello_hiho[16]
)
# contexts["a2"] == "1" sU
self.mora_hiho_4 = Mora(
consonant=self.phonemes_hello_hiho[17], vowel=self.phonemes_hello_hiho[18]
)
def assert_phonemes(self, mora: Mora, mora_str: str) -> None:
self.assertEqual(
"".join([phoneme.phoneme for phoneme in mora.phonemes]), mora_str
)
def assert_labels(self, mora: Mora, label_start: int, label_end: int) -> None:
self.assertEqual(mora.labels, self.test_case_hello_hiho[label_start:label_end])
def test_phonemes(self) -> None:
self.assert_phonemes(self.mora_hello_1, "ko")
self.assert_phonemes(self.mora_hello_2, "N")
self.assert_phonemes(self.mora_hello_3, "ni")
self.assert_phonemes(self.mora_hello_4, "chi")
self.assert_phonemes(self.mora_hello_5, "wa")
self.assert_phonemes(self.mora_hiho_1, "hi")
self.assert_phonemes(self.mora_hiho_2, "ho")
self.assert_phonemes(self.mora_hiho_3, "de")
self.assert_phonemes(self.mora_hiho_4, "sU")
def test_labels(self) -> None:
self.assert_labels(self.mora_hello_1, 1, 3)
self.assert_labels(self.mora_hello_2, 3, 4)
self.assert_labels(self.mora_hello_3, 4, 6)
self.assert_labels(self.mora_hello_4, 6, 8)
self.assert_labels(self.mora_hello_5, 8, 10)
self.assert_labels(self.mora_hiho_1, 11, 13)
self.assert_labels(self.mora_hiho_2, 13, 15)
self.assert_labels(self.mora_hiho_3, 15, 17)
self.assert_labels(self.mora_hiho_4, 17, 19)
def test_set_context(self):
# ๅ€คใ‚’ๆ›ธใๆ›ใˆใ‚‹ใฎใงใ€ไป–ใฎใƒ†ใ‚นใƒˆใซๅฝฑ้Ÿฟใ‚’ๅ‡บใ•ใชใ„ใŸใ‚ใซdeepcopyใ™ใ‚‹
mora_hello_1 = deepcopy(self.mora_hello_1)
# phonemeใซใ‚ใŸใ‚‹"p3"ใ‚’ๆ›ธใๆ›ใˆใ‚‹
mora_hello_1.set_context("p3", "a")
self.assert_phonemes(mora_hello_1, "aa")
class TestAccentPhrase(TestBasePhonemes):
def setUp(self) -> None:
super().setUp()
# TODO: ValueErrorใ‚’ๅใไฝœ็‚บ็š„ใงใฏใชใ„่‡ช็„ถใชไพ‹ใฎๆจก็ดข
# ๅญ˜ๅœจใ—ใชใ„ใชใ‚‰ๆ”พ็ฝฎใงใ‚ˆใ„
self.accent_phrase_hello = AccentPhrase.from_phonemes(
self.phonemes_hello_hiho[1:10]
)
self.accent_phrase_hiho = AccentPhrase.from_phonemes(
self.phonemes_hello_hiho[11:19]
)
def test_accent(self):
self.assertEqual(self.accent_phrase_hello.accent, 5)
self.assertEqual(self.accent_phrase_hiho.accent, 1)
def test_set_context(self):
accent_phrase_hello = deepcopy(self.accent_phrase_hello)
# phonemeใซใ‚ใŸใ‚‹"p3"ใ‚’ๆ›ธใๆ›ใˆใ‚‹
accent_phrase_hello.set_context("p3", "a")
self.assertEqual(
"".join([phoneme.phoneme for phoneme in accent_phrase_hello.phonemes]),
"aaaaaaaaa",
)
def test_phonemes(self):
self.assertEqual(
" ".join(
[phoneme.phoneme for phoneme in self.accent_phrase_hello.phonemes]
),
"k o N n i ch i w a",
)
self.assertEqual(
" ".join([phoneme.phoneme for phoneme in self.accent_phrase_hiho.phonemes]),
"h i h o d e s U",
)
def test_labels(self):
self.assertEqual(
self.accent_phrase_hello.labels, self.test_case_hello_hiho[1:10]
)
self.assertEqual(
self.accent_phrase_hiho.labels, self.test_case_hello_hiho[11:19]
)
def test_merge(self):
# ใ€Œใ“ใ‚“ใซใกใฏใƒ’ใƒ›ใงใ™ใ€
# ่ชญ็‚นใ‚’็„กใใ—ใŸใ‚‚ใฎใจๅŒ็ญ‰
merged_accent_phrase = self.accent_phrase_hello.merge(self.accent_phrase_hiho)
self.assertEqual(merged_accent_phrase.accent, 5)
self.assertEqual(
" ".join([phoneme.phoneme for phoneme in merged_accent_phrase.phonemes]),
"k o N n i ch i w a h i h o d e s U",
)
self.assertEqual(
merged_accent_phrase.labels,
self.test_case_hello_hiho[1:10] + self.test_case_hello_hiho[11:19],
)
class TestBreathGroup(TestBasePhonemes):
def setUp(self) -> None:
super().setUp()
self.breath_group_hello = BreathGroup.from_phonemes(
self.phonemes_hello_hiho[1:10]
)
self.breath_group_hiho = BreathGroup.from_phonemes(
self.phonemes_hello_hiho[11:19]
)
def test_set_context(self):
# ๅ€คใ‚’ๆ›ธใๆ›ใˆใ‚‹ใฎใงใ€ไป–ใฎใƒ†ใ‚นใƒˆใซๅฝฑ้Ÿฟใ‚’ๅ‡บใ•ใชใ„ใŸใ‚ใซdeepcopyใ™ใ‚‹
breath_group_hello = deepcopy(self.breath_group_hello)
# phonemeใซใ‚ใŸใ‚‹"p3"ใ‚’ๆ›ธใๆ›ใˆใ‚‹
breath_group_hello.set_context("p3", "a")
self.assertEqual(
"".join([phoneme.phoneme for phoneme in breath_group_hello.phonemes]),
"aaaaaaaaa",
)
def test_phonemes(self):
self.assertEqual(
" ".join([phoneme.phoneme for phoneme in self.breath_group_hello.phonemes]),
"k o N n i ch i w a",
)
self.assertEqual(
" ".join([phoneme.phoneme for phoneme in self.breath_group_hiho.phonemes]),
"h i h o d e s U",
)
def test_labels(self):
self.assertEqual(
self.breath_group_hello.labels, self.test_case_hello_hiho[1:10]
)
self.assertEqual(
self.breath_group_hiho.labels, self.test_case_hello_hiho[11:19]
)
class TestUtterance(TestBasePhonemes):
def setUp(self) -> None:
super().setUp()
self.utterance_hello_hiho = Utterance.from_phonemes(self.phonemes_hello_hiho)
def test_phonemes(self):
self.assertEqual(
" ".join(
[phoneme.phoneme for phoneme in self.utterance_hello_hiho.phonemes]
),
"sil k o N n i ch i w a pau h i h o d e s U sil",
)
changed_utterance = Utterance.from_phonemes(self.utterance_hello_hiho.phonemes)
self.assertEqual(len(changed_utterance.breath_groups), 2)
accent_phrases = list(
chain.from_iterable(
breath_group.accent_phrases
for breath_group in changed_utterance.breath_groups
)
)
for prev, cent, post in zip(
[None] + accent_phrases[:-1],
accent_phrases,
accent_phrases[1:] + [None],
):
mora_num = len(cent.moras)
accent = cent.accent
if prev is not None:
for phoneme in prev.phonemes:
self.assertEqual(phoneme.contexts["g1"], str(mora_num))
self.assertEqual(phoneme.contexts["g2"], str(accent))
if post is not None:
for phoneme in post.phonemes:
self.assertEqual(phoneme.contexts["e1"], str(mora_num))
self.assertEqual(phoneme.contexts["e2"], str(accent))
for phoneme in cent.phonemes:
self.assertEqual(
phoneme.contexts["k2"],
str(
sum(
[
len(breath_group.accent_phrases)
for breath_group in changed_utterance.breath_groups
]
)
),
)
for prev, cent, post in zip(
[None] + changed_utterance.breath_groups[:-1],
changed_utterance.breath_groups,
changed_utterance.breath_groups[1:] + [None],
):
accent_phrase_num = len(cent.accent_phrases)
if prev is not None:
for phoneme in prev.phonemes:
self.assertEqual(phoneme.contexts["j1"], str(accent_phrase_num))
if post is not None:
for phoneme in post.phonemes:
self.assertEqual(phoneme.contexts["h1"], str(accent_phrase_num))
for phoneme in cent.phonemes:
self.assertEqual(phoneme.contexts["i1"], str(accent_phrase_num))
self.assertEqual(
phoneme.contexts["i5"],
str(accent_phrases.index(cent.accent_phrases[0]) + 1),
)
self.assertEqual(
phoneme.contexts["i6"],
str(
len(accent_phrases)
- accent_phrases.index(cent.accent_phrases[0])
),
)
def test_labels(self):
self.assertEqual(self.utterance_hello_hiho.labels, self.test_case_hello_hiho)