Upload VerbaLex_voice.py
Browse files- VerbaLex_voice.py +134 -0
VerbaLex_voice.py
ADDED
@@ -0,0 +1,134 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import csv
|
2 |
+
import os
|
3 |
+
|
4 |
+
import datasets
|
5 |
+
from tqdm import tqdm
|
6 |
+
|
7 |
+
from VerbaLex_Voice.accents import ACCENTS
|
8 |
+
from VerbaLex_Voice.release_stats import STATS
|
9 |
+
|
10 |
+
_HOMEPAGE = "https://huggingface.co/datasets/RitchieP/VerbaLex_voice"
|
11 |
+
|
12 |
+
_LICENSE = "https://choosealicense.com/licenses/apache-2.0/"
|
13 |
+
|
14 |
+
_BASE_URL = "https://huggingface.co/datasets/RitchieP/VerbaLex_voice/tree/main"
|
15 |
+
|
16 |
+
_AUDIO_URL = _BASE_URL + "audio/{accent}/{split}/{accent}_{split}.tar"
|
17 |
+
|
18 |
+
_TRANSCRIPT_URL = _BASE_URL + "transcript/{accent}/{split}.tsv"
|
19 |
+
|
20 |
+
_CITATION = """\
|
21 |
+
"""
|
22 |
+
|
23 |
+
|
24 |
+
class VerbaLexVoiceConfig(datasets.BuilderConfig):
|
25 |
+
def __init__(self, name, version, **kwargs):
|
26 |
+
self.accent = kwargs.pop("accent", None)
|
27 |
+
self.num_speakers = kwargs.pop("num_speakers", None)
|
28 |
+
self.num_files = kwargs.pop("num_clips", None)
|
29 |
+
description = (
|
30 |
+
f"VerbaLex Voice english speech-to-text dataset in {self.accent} accent."
|
31 |
+
)
|
32 |
+
|
33 |
+
super(VerbaLexVoiceConfig, self).__init__(
|
34 |
+
name=name,
|
35 |
+
version=datasets.Version(version),
|
36 |
+
description=description,
|
37 |
+
**kwargs,
|
38 |
+
)
|
39 |
+
|
40 |
+
|
41 |
+
class VerbaLexVoiceDataset(datasets.GeneratorBasedBuilder):
|
42 |
+
"""
|
43 |
+
VerbaLex is a dataset containing different English accents from non-native English speakers.
|
44 |
+
This dataset is created directly from the L2-Arctic dataset.
|
45 |
+
"""
|
46 |
+
BUILDER_CONFIGS = [
|
47 |
+
VerbaLexVoiceConfig(
|
48 |
+
name=accent,
|
49 |
+
version=STATS["version"],
|
50 |
+
accent=ACCENTS[accent],
|
51 |
+
num_speakers=accent_stats["numOfSpeaker"],
|
52 |
+
num_files=accent_stats["numOfWavFiles"]
|
53 |
+
)
|
54 |
+
for accent, accent_stats in STATS["accents"].items()
|
55 |
+
]
|
56 |
+
|
57 |
+
DEFAULT_CONFIG_NAME = "all"
|
58 |
+
|
59 |
+
def _info(self):
|
60 |
+
return datasets.DatasetInfo(
|
61 |
+
description=(
|
62 |
+
"VerbaLex Voice is a speech dataset focusing on accented English speech."
|
63 |
+
"It specifically targets speeches from speakers that is a non-native English speaker."
|
64 |
+
),
|
65 |
+
features=datasets.Features(
|
66 |
+
{
|
67 |
+
"path": datasets.Value("string"),
|
68 |
+
"accent": datasets.Value("string"),
|
69 |
+
"sentence": datasets.Value("string"),
|
70 |
+
"audio": datasets.Audio(sampling_rate=44_100)
|
71 |
+
}
|
72 |
+
),
|
73 |
+
supervised_keys=None,
|
74 |
+
homepage=_HOMEPAGE,
|
75 |
+
license=_LICENSE,
|
76 |
+
citation=_CITATION
|
77 |
+
)
|
78 |
+
|
79 |
+
def _split_generators(self, dl_manager):
|
80 |
+
"""Returns SplitGenerators"""
|
81 |
+
accent = self.config.name
|
82 |
+
|
83 |
+
splits = ("train", "test")
|
84 |
+
audio_urls = {}
|
85 |
+
for split in splits:
|
86 |
+
audio_urls[split] = _AUDIO_URL.format(accent=accent, split=split)
|
87 |
+
archive_paths = dl_manager.download(audio_urls)
|
88 |
+
local_extracted_archive_paths = dl_manager.extract(archive_paths) if not dl_manager.is_streaming else {}
|
89 |
+
|
90 |
+
meta_urls = {split: _TRANSCRIPT_URL.format(accent=accent, split=split) for split in splits}
|
91 |
+
meta_paths = dl_manager.download_and_extract(meta_urls)
|
92 |
+
|
93 |
+
split_names = {
|
94 |
+
"train": datasets.Split.TRAIN,
|
95 |
+
"test": datasets.Split.TEST
|
96 |
+
}
|
97 |
+
split_generators = []
|
98 |
+
for split in splits:
|
99 |
+
split_generators.append(
|
100 |
+
datasets.SplitGenerator(
|
101 |
+
name=split_names.get(split, split),
|
102 |
+
gen_kwargs={
|
103 |
+
"local_extracted_archive_paths": local_extracted_archive_paths.get(split),
|
104 |
+
"archives": [dl_manager.iter_archive(path) for path in archive_paths.get(split)],
|
105 |
+
"meta_path": meta_paths[split]
|
106 |
+
}
|
107 |
+
)
|
108 |
+
)
|
109 |
+
|
110 |
+
return split_generators
|
111 |
+
|
112 |
+
def _generate_examples(self, local_extracted_archive_paths, archives, meta_path):
|
113 |
+
data_fields = list(self._info().features.keys())
|
114 |
+
metadata = {}
|
115 |
+
with open(meta_path, encoding="UTF-8") as f:
|
116 |
+
reader = csv.DictReader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
|
117 |
+
for row in tqdm(reader, desc="Reading metadata..."):
|
118 |
+
if not row["path"].endswith(".wav"):
|
119 |
+
row["path"] += ".wav"
|
120 |
+
for field in data_fields:
|
121 |
+
if field not in row:
|
122 |
+
row[field] = ""
|
123 |
+
metadata[row["path"]] = row
|
124 |
+
|
125 |
+
for i, audio_archive in enumerate(archives):
|
126 |
+
for path, file in audio_archive:
|
127 |
+
_, filename = os.path.split(path)
|
128 |
+
if filename in metadata:
|
129 |
+
result = dict(metadata[filename])
|
130 |
+
path = os.path.join(local_extracted_archive_paths[i],
|
131 |
+
path) if local_extracted_archive_paths else path
|
132 |
+
result["audio"] = {"path": path, "bytes": file.read()}
|
133 |
+
result["path"] = path
|
134 |
+
yield path, result
|