File size: 9,443 Bytes
cf15793
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
# coding=utf-8
import csv
import json
import os
from pathlib import Path
from typing import Dict, List, Tuple

import datasets

from seacrowd.utils import schemas
from seacrowd.utils.configs import SEACrowdConfig
from seacrowd.utils.constants import Licenses, Tasks

_CITATION = """\
@inproceedings{commonvoice:2020,
  author = {Ardila, R. and Branson, M. and Davis, K. and Henretty, M. and Kohler, M. and Meyer, J. and Morais, R. and Saunders, L. and Tyers, F. M. and Weber, G.},
  title = {Common Voice: A Massively-Multilingual Speech Corpus},
  booktitle = {Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020)},
  pages = {4211--4215},
  year = 2020
}
"""

_DATASETNAME = "commonvoice_120"

_DESCRIPTION = """\
The Common Mozilla Voice dataset consists of a unique MP3 and corresponding text file.
Many of the 26119 recorded hours in the dataset also include demographic metadata like age, sex, and accent that can help improve the accuracy of speech recognition engines.
The dataset currently consists of 17127 validated hours in 104 languages, but more voices and languages are always added.

Before using this dataloader, please accept the acknowledgement at https://huggingface.co/datasets/mozilla-foundation/common_voice_12_0 and use huggingface-cli login for authentication
"""

_HOMEPAGE = "https://commonvoice.mozilla.org/en/datasets"

_LANGUAGES = ["cnh", "ind", "tha", "vie"]
_LANG_TO_CVLANG = {"cnh": "cnh", "ind": "id", "tha": "th", "vie": "vi"}

_AGE_TO_INT = {"": None, "teens": 10, "twenties": 20, "thirties": 30, "fourties": 40, "fifties": 50, "sixties": 60, "seventies": 70, "eighties": 80}

_LICENSE = Licenses.CC0_1_0.value

# Note: the dataset is gated in HuggingFace. It's public after providing access token
_LOCAL = False

_COMMONVOICE_URL_TEMPLATE = "https://huggingface.co/datasets/mozilla-foundation/common_voice_12_0/resolve/main/"
_URLS = {"audio": _COMMONVOICE_URL_TEMPLATE + "audio/{lang}/{split}/{lang}_{split}_{shard_idx}.tar", "transcript": _COMMONVOICE_URL_TEMPLATE + "transcript/{lang}/{split}.tsv", "n_shards": _COMMONVOICE_URL_TEMPLATE + "n_shards.json"}

_SUPPORTED_TASKS = [Tasks.SPEECH_RECOGNITION, Tasks.TEXT_TO_SPEECH]

_SOURCE_VERSION = "1.0.0"

_SEACROWD_VERSION = "2024.06.20"


class Commonvoice120(datasets.GeneratorBasedBuilder):
    """This is the dataloader for CommonVoice 12.0 Mozilla"""

    SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
    SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)

    BUILDER_CONFIGS = (
        *[
            SEACrowdConfig(
                name=f"{_DATASETNAME}_{lang}{'_' if lang else ''}source",
                version=datasets.Version(_SOURCE_VERSION),
                description=f"{_DATASETNAME} source schema for {lang}",
                schema="source",
                subset_id=f"{_DATASETNAME}{'_' if lang else ''}{lang}",
            )
            for lang in ["", *_LANGUAGES]
        ],
        *[
            SEACrowdConfig(
                name=f"{_DATASETNAME}_{lang}{'_' if lang else ''}seacrowd_sptext",
                version=datasets.Version(_SEACROWD_VERSION),
                description=f"{_DATASETNAME} SEACrowd schema for {lang}",
                schema="seacrowd_sptext",
                subset_id=f"{_DATASETNAME}{'_' if lang else ''}{lang}",
            )
            for lang in ["", *_LANGUAGES]
        ],
    )

    DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"

    def _info(self) -> datasets.DatasetInfo:
        if self.config.schema == "source":
            features = datasets.Features(
                {
                    "client_id": datasets.Value("string"),
                    "path": datasets.Value("string"),
                    "audio": datasets.features.Audio(sampling_rate=48_000),
                    "sentence": datasets.Value("string"),
                    "up_votes": datasets.Value("int64"),
                    "down_votes": datasets.Value("int64"),
                    "age": datasets.Value("string"),
                    "gender": datasets.Value("string"),
                    "accent": datasets.Value("string"),
                    "locale": datasets.Value("string"),
                    "segment": datasets.Value("string"),
                }
            )
        elif self.config.schema == "seacrowd_sptext":
            features = schemas.speech_text_features

        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
        lang_code = self.config.subset_id.split("_")[-1]
        languages = [_LANG_TO_CVLANG.get(lang, lang) for lang in (_LANGUAGES if lang_code == "120" else [lang_code])]
        n_shards_path = dl_manager.download_and_extract(_URLS["n_shards"])
        with open(n_shards_path, encoding="utf-8") as f:
            n_shards = json.load(f)

        audio_urls = {}
        meta_urls = {}
        splits = ("train", "dev", "test")
        for split in splits:
            audio_urls[split] = [_URLS["audio"].format(lang=lang, split=split, shard_idx=i) for lang in languages for i in range(n_shards[lang][split])]
            meta_urls[split] = [_URLS["transcript"].format(lang=lang, split=split) for lang in languages]
        archive_paths = dl_manager.download(audio_urls)
        local_extracted_archive_paths = dl_manager.extract(archive_paths)
        meta_paths = dl_manager.download_and_extract(meta_urls)

        split_names = {
            "train": datasets.Split.TRAIN,
            "dev": datasets.Split.VALIDATION,
            "test": datasets.Split.TEST,
        }
        return [
            datasets.SplitGenerator(
                name=split_names.get(split, split),
                gen_kwargs={
                    "local_extracted_archive_paths": local_extracted_archive_paths.get(split),
                    "audio_archives": [dl_manager.iter_archive(path) for path in archive_paths.get(split)],
                    "meta_paths": meta_paths[split],
                    "split": "train",
                },
            )
            for split in splits
        ]

    def _generate_examples(self, local_extracted_archive_paths: [Path], audio_archives: [Path], meta_paths: [Path], split: str) -> Tuple[int, Dict]:
        data_fields = list(self._info().features.keys())
        metadata = {}
        for meta_path in meta_paths:
            with open(meta_path, encoding="utf-8") as f:
                reader = csv.DictReader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
                for row in reader:
                    if not row["path"].endswith(".mp3"):
                        row["path"] += ".mp3"
                    if "accents" in row:
                        row["accent"] = row["accents"]
                        del row["accents"]
                    for field in data_fields:
                        if field not in row:
                            row[field] = ""
                    metadata[row["path"]] = row

        if self.config.schema == "source":
            for i, audio_archive in enumerate(audio_archives):
                for path, file in audio_archive:
                    _, filename = os.path.split(path)
                    if filename in metadata:
                        src_result = dict(metadata[filename])
                        path = os.path.join(local_extracted_archive_paths[i], path)
                        result = {
                            "client_id": src_result["client_id"],
                            "path": path,
                            "audio": {"path": path, "bytes": file.read()},
                            "sentence": src_result["sentence"],
                            "up_votes": src_result["up_votes"],
                            "down_votes": src_result["down_votes"],
                            "age": src_result["age"],
                            "gender": src_result["gender"],
                            "accent": src_result["accent"],
                            "locale": src_result["locale"],
                            "segment": src_result["segment"],
                        }
                        yield path, result

        elif self.config.schema == "seacrowd_sptext":
            for i, audio_archive in enumerate(audio_archives):
                for path, file in audio_archive:
                    _, filename = os.path.split(path)
                    if filename in metadata:
                        src_result = dict(metadata[filename])
                        # set the audio feature and the path to the extracted file
                        path = os.path.join(local_extracted_archive_paths[i], path)
                        result = {
                            "id": src_result["path"].replace(".mp3", ""),
                            "path": path,
                            "audio": {"path": path, "bytes": file.read()},
                            "text": src_result["sentence"],
                            "speaker_id": src_result["client_id"],
                            "metadata": {
                                "speaker_age": _AGE_TO_INT[src_result["age"]],
                                "speaker_gender": src_result["gender"],
                            },
                        }
                        yield path, result