forked from MontrealCorpusTools/Montreal-Forced-Aligner
-
Notifications
You must be signed in to change notification settings - Fork 0
/
test_config.py
190 lines (164 loc) · 5.81 KB
/
test_config.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
import os
import pytest
from montreal_forced_aligner.acoustic_modeling import (
LdaTrainer,
MonophoneTrainer,
SatTrainer,
TrainableAligner,
TriphoneTrainer,
)
from montreal_forced_aligner.alignment import PretrainedAligner
from montreal_forced_aligner.exceptions import ConfigError
from montreal_forced_aligner.ivector.trainer import TrainableIvectorExtractor
def test_monophone_config(basic_corpus_dir, basic_dict_path, temp_dir):
am_trainer = TrainableAligner(
corpus_directory=basic_corpus_dir,
dictionary_path=basic_dict_path,
temporary_directory=temp_dir,
)
config = MonophoneTrainer(identifier="mono", worker=am_trainer)
config.compute_calculated_properties()
assert config.realignment_iterations == [
0,
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
12,
14,
16,
18,
20,
23,
26,
29,
32,
35,
38,
]
am_trainer.cleanup()
def test_triphone_config(basic_corpus_dir, basic_dict_path, temp_dir):
am_trainer = TrainableAligner(
corpus_directory=basic_corpus_dir,
dictionary_path=basic_dict_path,
temporary_directory=temp_dir,
)
config = TriphoneTrainer(identifier="tri", worker=am_trainer)
config.compute_calculated_properties()
assert config.realignment_iterations == [10, 20, 30]
am_trainer.cleanup()
def test_lda_mllt_config(basic_corpus_dir, basic_dict_path, temp_dir):
am_trainer = TrainableAligner(
corpus_directory=basic_corpus_dir,
dictionary_path=basic_dict_path,
temporary_directory=temp_dir,
)
assert am_trainer.beam == 10
assert am_trainer.retry_beam == 40
assert am_trainer.align_options["beam"] == 10
assert am_trainer.align_options["retry_beam"] == 40
config = LdaTrainer(identifier="lda", worker=am_trainer)
config.compute_calculated_properties()
assert config.mllt_iterations == [2, 4, 6, 12]
am_trainer.cleanup()
def test_load_align(
config_directory,
basic_corpus_dir,
basic_dict_path,
temp_dir,
english_acoustic_model,
mono_align_config_path,
):
params = PretrainedAligner.parse_parameters(mono_align_config_path)
aligner = PretrainedAligner(
acoustic_model_path=english_acoustic_model,
corpus_directory=basic_corpus_dir,
dictionary_path=basic_dict_path,
temporary_directory=temp_dir,
**params
)
assert params["beam"] == 100
assert params["retry_beam"] == 400
assert aligner.beam == 100
assert aligner.retry_beam == 400
assert aligner.align_options["beam"] == 100
assert aligner.align_options["retry_beam"] == 400
aligner.cleanup()
path = os.path.join(config_directory, "bad_align_config.yaml")
params = PretrainedAligner.parse_parameters(path)
print(params)
aligner = PretrainedAligner(
acoustic_model_path=english_acoustic_model,
corpus_directory=basic_corpus_dir,
dictionary_path=basic_dict_path,
temporary_directory=temp_dir,
**params
)
assert aligner.beam == 10
assert aligner.retry_beam == 40
aligner.cleanup()
def test_load_basic_train(basic_corpus_dir, basic_dict_path, temp_dir, basic_train_config_path):
params = TrainableAligner.parse_parameters(basic_train_config_path)
am_trainer = TrainableAligner(
corpus_directory=basic_corpus_dir,
dictionary_path=basic_dict_path,
temporary_directory=temp_dir,
**params
)
assert am_trainer.beam == 100
assert am_trainer.retry_beam == 400
assert am_trainer.align_options["beam"] == 100
assert am_trainer.align_options["retry_beam"] == 400
for trainer in am_trainer.training_configs.values():
assert trainer.beam == 100
assert trainer.retry_beam == 400
assert trainer.align_options["beam"] == 100
assert trainer.align_options["retry_beam"] == 400
am_trainer.cleanup()
def test_load_mono_train(basic_corpus_dir, basic_dict_path, temp_dir, mono_train_config_path):
params = TrainableAligner.parse_parameters(mono_train_config_path)
am_trainer = TrainableAligner(
corpus_directory=basic_corpus_dir,
dictionary_path=basic_dict_path,
temporary_directory=temp_dir,
**params
)
for t in am_trainer.training_configs.values():
assert not t.use_mp
assert t.use_energy
assert not am_trainer.use_mp
assert am_trainer.use_energy
am_trainer.cleanup()
def test_load_ivector_train(basic_corpus_dir, temp_dir, train_ivector_config_path):
params = TrainableIvectorExtractor.parse_parameters(train_ivector_config_path)
trainer = TrainableIvectorExtractor(
corpus_directory=basic_corpus_dir, temporary_directory=temp_dir, **params
)
for t in trainer.training_configs.values():
assert not t.use_mp
assert t.use_energy
assert not trainer.use_mp
trainer.cleanup()
def test_load(basic_corpus_dir, basic_dict_path, temp_dir, config_directory):
path = os.path.join(config_directory, "basic_train_config.yaml")
params = TrainableAligner.parse_parameters(path)
am_trainer = TrainableAligner(
corpus_directory=basic_corpus_dir,
dictionary_path=basic_dict_path,
temporary_directory=temp_dir,
**params
)
assert len(am_trainer.training_configs) == 4
assert isinstance(am_trainer.training_configs["monophone"], MonophoneTrainer)
assert isinstance(am_trainer.training_configs["triphone"], TriphoneTrainer)
assert isinstance(am_trainer.training_configs[am_trainer.final_identifier], SatTrainer)
path = os.path.join(config_directory, "out_of_order_config.yaml")
with pytest.raises(ConfigError):
params = TrainableAligner.parse_parameters(path)
am_trainer.cleanup()