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
|
import os
from torchaudio.datasets import speechcommands
from torchaudio_unittest.common_utils import (
TempDirMixin,
TorchaudioTestCase,
get_whitenoise,
normalize_wav,
save_wav,
)
LABELS = [
"bed",
"bird",
"cat",
"dog",
"down",
"eight",
"five",
"follow",
"forward",
"four",
"go",
"happy",
"house",
"learn",
"left",
"marvin",
"nine",
"no",
"off",
"on",
"one",
"right",
"seven",
"sheila",
"six",
"stop",
"three",
"tree",
"two",
"up",
"visual",
"wow",
"yes",
"zero",
]
class TestSpeechCommands(TempDirMixin, TorchaudioTestCase):
backend = "default"
root_dir = None
samples = []
@classmethod
def setUp(cls):
cls.root_dir = cls.get_base_temp_dir()
dataset_dir = os.path.join(
cls.root_dir, speechcommands.FOLDER_IN_ARCHIVE, speechcommands.URL
)
os.makedirs(dataset_dir, exist_ok=True)
sample_rate = 16000 # 16kHz sample rate
seed = 0
for label in LABELS:
path = os.path.join(dataset_dir, label)
os.makedirs(path, exist_ok=True)
for j in range(2):
# generate hash ID for speaker
speaker = "{:08x}".format(j)
for utterance in range(3):
filename = f"{speaker}{speechcommands.HASH_DIVIDER}{utterance}.wav"
file_path = os.path.join(path, filename)
seed += 1
data = get_whitenoise(
sample_rate=sample_rate,
duration=0.01,
n_channels=1,
dtype="int16",
seed=seed,
)
save_wav(file_path, data, sample_rate)
sample = (
normalize_wav(data),
sample_rate,
label,
speaker,
utterance,
)
cls.samples.append(sample)
def testSpeechCommands(self):
dataset = speechcommands.SPEECHCOMMANDS(self.root_dir)
print(dataset._path)
num_samples = 0
for i, (data, sample_rate, label, speaker_id, utterance_number) in enumerate(
dataset
):
self.assertEqual(data, self.samples[i][0], atol=5e-5, rtol=1e-8)
assert sample_rate == self.samples[i][1]
assert label == self.samples[i][2]
assert speaker_id == self.samples[i][3]
assert utterance_number == self.samples[i][4]
num_samples += 1
assert num_samples == len(self.samples)
|
