File size: 6,385 Bytes
9d61c9b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
import unittest

import torch

from training.preprocess.wav2vec_aligner import Wav2VecAligner


class TestWav2VecAligner(unittest.TestCase):
    def setUp(self):
        self.model = Wav2VecAligner()
        self.text = "I HAD THAT CURIOSITY BESIDE ME AT THIS MOMENT"
        self.wav_path = "./mocks/audio_example.wav"

    def test_load_audio(self):
        _, sample_rate = self.model.load_audio(self.wav_path)

        self.assertEqual(sample_rate, 16_000)

        with self.assertRaises(FileNotFoundError):
            self.model.load_audio("./nonexistent/path.wav")

    def test_encode(self):
        tokens = self.model.encode(self.text)

        torch.testing.assert_close(
            tokens,
            torch.tensor(
                [
                    [
                        10,
                        4,
                        11,
                        7,
                        14,
                        4,
                        6,
                        11,
                        7,
                        6,
                        4,
                        19,
                        16,
                        13,
                        10,
                        8,
                        12,
                        10,
                        6,
                        22,
                        4,
                        24,
                        5,
                        12,
                        10,
                        14,
                        5,
                        4,
                        17,
                        5,
                        4,
                        7,
                        6,
                        4,
                        6,
                        11,
                        10,
                        12,
                        4,
                        17,
                        8,
                        17,
                        5,
                        9,
                        6,
                    ],
                ],
            ),
        )

    def test_decode(self):
        transcript = self.model.decode(
            [
                [
                    10,
                    4,
                    11,
                    7,
                    14,
                    4,
                    6,
                    11,
                    7,
                    6,
                    4,
                    19,
                    16,
                    13,
                    10,
                    8,
                    12,
                    10,
                    6,
                    22,
                    4,
                    24,
                    5,
                    12,
                    10,
                    14,
                    5,
                    4,
                    17,
                    5,
                    4,
                    7,
                    6,
                    4,
                    6,
                    11,
                    10,
                    12,
                    4,
                    17,
                    8,
                    17,
                    5,
                    9,
                    6,
                ],
            ],
        )

        self.assertEqual(transcript, self.text)

    def test_align_single_sample(self):
        audio_input, _ = self.model.load_audio(self.wav_path)
        emissions, tokens, transcript = self.model.align_single_sample(
            audio_input, self.text,
        )

        self.assertEqual(emissions.shape, torch.Size([169, 32]))

        self.assertEqual(
            len(tokens),
            47,
        )

        self.assertEqual(transcript, "|I|HAD|THAT|CURIOSITY|BESIDE|ME|AT|THIS|MOMENT|")

    def test_get_trellis(self):
        audio_input, _ = self.model.load_audio(self.wav_path)
        emissions, tokens, _ = self.model.align_single_sample(audio_input, self.text)
        trellis = self.model.get_trellis(emissions, tokens)

        self.assertEqual(emissions.shape, torch.Size([169, 32]))
        self.assertEqual(len(tokens), 47)

        # Add assertions here based on the expected behavior of get_trellis
        self.assertIsInstance(trellis, torch.Tensor)
        self.assertEqual(trellis.shape, torch.Size([169, 47]))

    def test_backtrack(self):
        audio_input, _ = self.model.load_audio(self.wav_path)
        emissions, tokens, _ = self.model.align_single_sample(audio_input, self.text)
        trellis = self.model.get_trellis(emissions, tokens)
        path = self.model.backtrack(trellis, emissions, tokens)

        # Add assertions here based on the expected behavior of backtrack
        self.assertIsInstance(path, list)
        self.assertEqual(len(path), 169)

    def test_merge_repeats(self):
        audio_input, _ = self.model.load_audio(self.wav_path)
        emissions, tokens, transcript = self.model.align_single_sample(
            audio_input, self.text,
        )
        trellis = self.model.get_trellis(emissions, tokens)
        path = self.model.backtrack(trellis, emissions, tokens)
        merged_path = self.model.merge_repeats(path, transcript)

        # Add assertions here based on the expected behavior of merge_repeats
        self.assertIsInstance(merged_path, list)
        self.assertEqual(len(merged_path), 47)

    def test_merge_words(self):
        audio_input, _ = self.model.load_audio(self.wav_path)
        emissions, tokens, transcript = self.model.align_single_sample(
            audio_input, self.text,
        )
        trellis = self.model.get_trellis(emissions, tokens)
        path = self.model.backtrack(trellis, emissions, tokens)
        merged_path = self.model.merge_repeats(path, transcript)
        merged_words = self.model.merge_words(merged_path)

        # Add assertions here based on the expected behavior of merge_words
        self.assertIsInstance(merged_words, list)
        self.assertEqual(len(merged_words), 9)

    def test_forward(self):
        result = self.model(self.wav_path, self.text)

        # self.assertEqual(result, expected_result)
        self.assertEqual(len(result), 9)

    def test_save_segments(self):
        # self.model.save_segments(self.wav_path, self.text, "./mocks/wav2vec_aligner/audio")
        self.assertEqual(True, True)


if __name__ == "__main__":
    unittest.main()