holylovenia
commited on
Upload kheng_info.py with huggingface_hub
Browse files- kheng_info.py +113 -0
kheng_info.py
ADDED
@@ -0,0 +1,113 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
|
3 |
+
from pathlib import Path
|
4 |
+
from typing import Dict, List, Tuple
|
5 |
+
|
6 |
+
import datasets
|
7 |
+
import pandas as pd
|
8 |
+
|
9 |
+
from seacrowd.utils import schemas
|
10 |
+
from seacrowd.utils.configs import SEACrowdConfig
|
11 |
+
from seacrowd.utils.constants import Licenses, Tasks
|
12 |
+
|
13 |
+
# no bibtex citation
|
14 |
+
_CITATION = ""
|
15 |
+
|
16 |
+
_DATASETNAME = "kheng_info"
|
17 |
+
|
18 |
+
_DESCRIPTION = """\
|
19 |
+
The Kheng.info Speech dataset was derived from recordings of Khmer words on the Khmer dictionary website kheng.info.
|
20 |
+
The recordings were recorded by a native Khmer speaker.
|
21 |
+
The recordings are short, generally ranging between 1 to 2 seconds only.
|
22 |
+
"""
|
23 |
+
|
24 |
+
_HOMEPAGE = "https://huggingface.co/datasets/seanghay/khmer_kheng_info_speech"
|
25 |
+
|
26 |
+
_LANGUAGES = ["khm"]
|
27 |
+
|
28 |
+
_LICENSE = Licenses.UNKNOWN.value
|
29 |
+
|
30 |
+
_LOCAL = False
|
31 |
+
|
32 |
+
_URLS = {
|
33 |
+
_DATASETNAME: "https://huggingface.co/datasets/seanghay/khmer_kheng_info_speech/resolve/main/data/train-00000-of-00001-4e7ad082a34164d1.parquet",
|
34 |
+
}
|
35 |
+
|
36 |
+
_SUPPORTED_TASKS = [Tasks.SPEECH_RECOGNITION]
|
37 |
+
|
38 |
+
_SOURCE_VERSION = "1.0.0"
|
39 |
+
|
40 |
+
_SEACROWD_VERSION = "2024.06.20"
|
41 |
+
|
42 |
+
|
43 |
+
class KhengInfoDataset(datasets.GeneratorBasedBuilder):
|
44 |
+
"""This is the Kheng.info Speech dataset, which wasderived from recordings on the Khmer dictionary website kheng.info"""
|
45 |
+
|
46 |
+
SOURCE_VERSION = datasets.Version(_SOURCE_VERSION)
|
47 |
+
SEACROWD_VERSION = datasets.Version(_SEACROWD_VERSION)
|
48 |
+
|
49 |
+
BUILDER_CONFIGS = [
|
50 |
+
SEACrowdConfig(
|
51 |
+
name=f"{_DATASETNAME}_source",
|
52 |
+
version=SOURCE_VERSION,
|
53 |
+
description=f"{_DATASETNAME} source schema",
|
54 |
+
schema="source",
|
55 |
+
subset_id=f"{_DATASETNAME}",
|
56 |
+
),
|
57 |
+
SEACrowdConfig(
|
58 |
+
name=f"{_DATASETNAME}_seacrowd_sptext",
|
59 |
+
version=SEACROWD_VERSION,
|
60 |
+
description=f"{_DATASETNAME} SEACrowd schema",
|
61 |
+
schema="seacrowd_sptext",
|
62 |
+
subset_id=f"{_DATASETNAME}",
|
63 |
+
),
|
64 |
+
]
|
65 |
+
|
66 |
+
DEFAULT_CONFIG_NAME = f"{_DATASETNAME}_source"
|
67 |
+
|
68 |
+
def _info(self) -> datasets.DatasetInfo:
|
69 |
+
if self.config.schema == "source":
|
70 |
+
features = datasets.Features({"word": datasets.Value("string"), "duration_ms": datasets.Value("int64"), "audio": datasets.Audio(sampling_rate=16_000)})
|
71 |
+
|
72 |
+
elif self.config.schema == "seacrowd_sptext":
|
73 |
+
features = schemas.speech_text_features
|
74 |
+
|
75 |
+
return datasets.DatasetInfo(
|
76 |
+
description=_DESCRIPTION,
|
77 |
+
features=features,
|
78 |
+
homepage=_HOMEPAGE,
|
79 |
+
license=_LICENSE,
|
80 |
+
citation=_CITATION,
|
81 |
+
)
|
82 |
+
|
83 |
+
def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
|
84 |
+
urls = _URLS[_DATASETNAME]
|
85 |
+
data_dir = dl_manager.download_and_extract(urls)
|
86 |
+
|
87 |
+
return [
|
88 |
+
datasets.SplitGenerator(
|
89 |
+
name=datasets.Split.TRAIN,
|
90 |
+
gen_kwargs={
|
91 |
+
"filepath": data_dir,
|
92 |
+
},
|
93 |
+
)
|
94 |
+
]
|
95 |
+
|
96 |
+
def _generate_examples(self, filepath: Path) -> Tuple[int, Dict]:
|
97 |
+
df = pd.read_parquet(filepath, engine="pyarrow")
|
98 |
+
if self.config.schema == "source":
|
99 |
+
for _id, row in df.iterrows():
|
100 |
+
yield _id, {"word": row["word"], "duration_ms": row["duration_ms"], "audio": row["audio"]}
|
101 |
+
elif self.config.schema == "seacrowd_sptext":
|
102 |
+
for _id, row in df.iterrows():
|
103 |
+
yield _id, {
|
104 |
+
"id": _id,
|
105 |
+
"path": row["audio"],
|
106 |
+
"audio": row["audio"],
|
107 |
+
"text": row["word"],
|
108 |
+
"speaker_id": None,
|
109 |
+
"metadata": {
|
110 |
+
"speaker_age": None,
|
111 |
+
"speaker_gender": None,
|
112 |
+
},
|
113 |
+
}
|