File size: 10,103 Bytes
7155eaf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8db4a65
8f6c4ad
7155eaf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2f1bff9
7155eaf
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
f3bf4e9
7155eaf
 
 
2f1bff9
 
 
 
 
 
 
 
 
 
7155eaf
2f1bff9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7155eaf
 
 
2f1bff9
 
 
 
 
 
7155eaf
 
 
 
2f1bff9
 
 
742365f
7155eaf
 
 
 
 
2f1bff9
 
 
742365f
7155eaf
 
 
 
 
 
 
2f1bff9
 
 
7155eaf
 
 
 
2f1bff9
 
 
7155eaf
 
 
 
2f1bff9
7155eaf
 
 
 
7374369
7155eaf
 
 
 
 
 
 
 
 
 
8db4a65
7155eaf
 
 
 
 
 
 
 
 
 
 
 
 
 
2f1bff9
 
 
 
7155eaf
2f1bff9
 
 
 
 
7155eaf
 
 
 
 
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
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
# coding=utf-8
# Copyright 2022 The TensorFlow Datasets Authors and the HuggingFace Datasets Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# Lint as: python3
"""Multilingual Librispeech automatic speech recognition dataset."""

import os

import datasets


_CITATION = """\
@article{Pratap2020MLSAL,
  title={MLS: A Large-Scale Multilingual Dataset for Speech Research},
  author={Vineel Pratap and Qiantong Xu and Anuroop Sriram and Gabriel Synnaeve and Ronan Collobert},
  journal={ArXiv},
  year={2020},
  volume={abs/2012.03411}
}
"""

_DESCRIPTION = """\
This is a streamable version of the Multilingual LibriSpeech (MLS) dataset. 
The data archives were restructured from the original ones from [OpenSLR](http://www.openslr.org/94) 
to make it easier to stream. 

MLS dataset is a large multilingual corpus suitable for speech research. 
The dataset is derived from read audiobooks from LibriVox and consists of 8 languages: 
English, German, Dutch, Spanish, French, Italian, Portuguese, Polish.
"""

_URL = "http://www.openslr.org/94"

_DL_URL_FORMAT = "data/mls_{name}"


class MultilingualLibrispeechConfig(datasets.BuilderConfig):
    """BuilderConfig for MultilingualLibrispeech."""

    def __init__(self, name, **kwargs):
        """
        Args:
          name: `string`, name of dataset config (=language)
          **kwargs: keyword arguments forwarded to super.
        """
        super(MultilingualLibrispeechConfig, self).__init__(
            version=datasets.Version("2.1.0", ""), name=name, **kwargs
        )
        # relative path to full data inside a repo (for example `data/mls_german`)
        self.data_root_url = _DL_URL_FORMAT.format(name=name)


class MultilingualLibrispeech(datasets.GeneratorBasedBuilder):
    """Multilingual Librispeech dataset."""

    BUILDER_CONFIGS = [
        MultilingualLibrispeechConfig(name="german", description="German LibriSpeech dataset"),
        MultilingualLibrispeechConfig(name="dutch", description="Dutch LibriSpeech dataset"),
        MultilingualLibrispeechConfig(name="french", description="French LibriSpeech dataset"),
        MultilingualLibrispeechConfig(name="spanish", description="Spanish LibriSpeech dataset"),
        MultilingualLibrispeechConfig(name="italian", description="Italian LibriSpeech dataset"),
        MultilingualLibrispeechConfig(name="portuguese", description="Portuguese LibriSpeech dataset"),
        MultilingualLibrispeechConfig(name="polish", description="Polish LibriSpeech dataset"),
    ]

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "file": datasets.Value("string"),
                    "audio": datasets.features.Audio(sampling_rate=16_000),
                    "text": datasets.Value("string"),
                    "speaker_id": datasets.Value("int64"),
                    "chapter_id": datasets.Value("int64"),
                    "id": datasets.Value("string"),
                }
            ),
            supervised_keys=("file", "text"),
            homepage=_URL,
            citation=_CITATION,
            task_templates=None,
        )

    def _split_generators(self, dl_manager):
        
        transcripts = dl_manager.download({
            "train": self.config.data_root_url + "/train/transcripts.txt",
            "dev": self.config.data_root_url + "/dev/transcripts.txt",
            "test": self.config.data_root_url + "/test/transcripts.txt",
        })

        # Download handles.txt files containing ids for limited supervision train sets
        limited_supervision_9h = dl_manager.download(
            [self.config.data_root_url + "/train/limited_supervision/9hr/handles.txt"],
        )
        # in our case of 1 hour limited supervision ("train.1h") there are always 6 subfolders like:
        # "limited_supervision/1h/0/handles.txt", "limited_supervision/1h/1/handles.txt", ...
        limited_supervision_1h = dl_manager.download([
            self.config.data_root_url + f"/train/limited_supervision/1hr/{i}/handles.txt" for i in range(6)
        ])
        
        # each split contains many .tar.gz archives with its audio files
        # audio_filenames.txt contains the names of these archives
        audio_filenames_paths = dl_manager.download({
            "train": self.config.data_root_url + "/train/audio_filenames.txt",
            "dev": self.config.data_root_url + "/dev/audio_filenames.txt",
            "test": self.config.data_root_url + "/test/audio_filenames.txt",
        })

        audio_archives = {}
        for split in audio_filenames_paths:
            with open(audio_filenames_paths[split], encoding="utf-8") as f:
                audio_filenames = [line.strip() for line in f.readlines()]
                audio_archives[split] = dl_manager.download([
                    self.config.data_root_url + "/" + split + "/audio/" + filename
                    for filename in audio_filenames
                ])

        # (Optional) In non-streaming mode, we can extract the archive locally to have actual local audio files:
        local_extracted_archives = dl_manager.extract(audio_archives) if not dl_manager.is_streaming else {}

        train_splits = [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={
                    "transcript_path": transcripts["train"],
                    "audio_archives": [dl_manager.iter_archive(archive) for archive in audio_archives["train"]],
                    "local_extracted_archive": local_extracted_archives.get("train"),
                }
            ),
            datasets.SplitGenerator(
                name="train.9h",
                gen_kwargs={
                    "transcript_path": transcripts["train"],
                    "audio_archives": [dl_manager.iter_archive(archive) for archive in audio_archives["train"]],
                    "local_extracted_archive": local_extracted_archives.get("train"),
                    "limited_ids_paths": tuple(limited_supervision_9h),
                },
            ),
            datasets.SplitGenerator(
                name="train.1h",
                gen_kwargs={
                    "transcript_path": transcripts["train"],
                    "audio_archives": [dl_manager.iter_archive(archive) for archive in audio_archives["train"]],
                    "local_extracted_archive": local_extracted_archives.get("train"),
                    "limited_ids_paths": tuple(limited_supervision_1h),
                },
            ),
        ]

        return train_splits + [
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION, gen_kwargs={
                    "transcript_path": transcripts["dev"],
                    "audio_archives": [dl_manager.iter_archive(archive) for archive in audio_archives["dev"]],
                    "local_extracted_archive": local_extracted_archives.get("dev"),
                }
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST, gen_kwargs={
                    "transcript_path": transcripts["test"],
                    "audio_archives": [dl_manager.iter_archive(archive) for archive in audio_archives["test"]],
                    "local_extracted_archive": local_extracted_archives.get("test"),
                }
            ),
        ]

    def _generate_examples(self, transcript_path, audio_archives, local_extracted_archive, limited_ids_paths=None):
        """Generate examples from a Multilingual LibriSpeech data dir."""
        transcripts = dict()
        with open(transcript_path, "r", encoding="utf-8") as file:
            for line in file:
                audio_id, transcript = line.strip().split("\t")
                transcripts[audio_id] = transcript

        limited_ids, limited_ids_archives_names = [], []
        if limited_ids_paths:
            for path in limited_ids_paths:
                with open(path, "r", encoding="utf-8") as file:
                    limited_ids.extend([line.strip() for line in file.readlines()])

            limited_ids = set(limited_ids)

        for archive_idx, audio_archive in enumerate(audio_archives):
            #  TODO: check that archive doesn't contain needed ids
            # if limited_ids and audio_archive not in limited_ids_archives_names:
            #     continue

            for audio_filename, file in audio_archive:
                speaker_id, chapter_id = audio_filename.split("_")[:2]
                speaker_id, chapter_id = int(speaker_id), int(chapter_id)
                audio_id = audio_filename.split(".flac")[0]
                audio_transcript = transcripts[audio_id]

                if limited_ids and audio_id not in limited_ids:
                    # this only can be true in limited supervision sets ("train.9h" and "train.1h")
                    continue

                local_audio_file_path = os.path.join(
                    local_extracted_archive[archive_idx], audio_filename
                ) if local_extracted_archive else None

                yield audio_filename, {
                    "file": local_audio_file_path,
                    "audio": {
                        "path": local_audio_file_path if local_audio_file_path else audio_filename,
                        "bytes": file.read()
                    },
                    "text": audio_transcript,
                    "speaker_id": speaker_id,
                    "chapter_id": chapter_id,
                    "id": audio_id
                }