cosmetics change
Browse files- librivox-indonesia.py +6 -7
librivox-indonesia.py
CHANGED
@@ -30,7 +30,7 @@ _HOMEPAGE = "https://huggingface.co/indonesian-nlp/librivox-indonesia"
|
|
30 |
|
31 |
_LICENSE = "https://creativecommons.org/publicdomain/zero/1.0/"
|
32 |
|
33 |
-
_DATA_URL = "https://huggingface.co/datasets/
|
34 |
|
35 |
|
36 |
class LibriVoxIndonesiaConfig(datasets.BuilderConfig):
|
@@ -41,13 +41,12 @@ class LibriVoxIndonesiaConfig(datasets.BuilderConfig):
|
|
41 |
self.release_date = kwargs.pop("release_date", None)
|
42 |
self.num_clips = kwargs.pop("num_clips", None)
|
43 |
self.num_speakers = kwargs.pop("num_speakers", None)
|
44 |
-
self.validated_hr = kwargs.pop("validated_hr", None)
|
45 |
self.total_hr = kwargs.pop("total_hr", None)
|
46 |
self.size_bytes = kwargs.pop("size_bytes", None)
|
47 |
self.size_human = size_str(self.size_bytes)
|
48 |
description = (
|
49 |
f"LibriVox-Indonesia speech to text dataset in {self.language} released on {self.release_date}. "
|
50 |
-
f"The dataset comprises {self.
|
51 |
)
|
52 |
super(LibriVoxIndonesiaConfig, self).__init__(
|
53 |
name=name,
|
@@ -105,13 +104,13 @@ class LibriVoxIndonesia(datasets.GeneratorBasedBuilder):
|
|
105 |
def _split_generators(self, dl_manager):
|
106 |
"""Returns SplitGenerators."""
|
107 |
dl_manager.download_config.ignore_url_params = True
|
108 |
-
|
109 |
local_extracted_archive = {}
|
110 |
metadata_path = {}
|
111 |
split_type = {"train": datasets.Split.TRAIN, "test": datasets.Split.TEST}
|
112 |
for split in split_type:
|
113 |
-
|
114 |
-
local_extracted_archive[split] = dl_manager.extract(
|
115 |
metadata_path[split] = dl_manager.download_and_extract(f"{_DATA_URL}/metadata_{split}.csv.gz")
|
116 |
path_to_clips = "librivox-indonesia"
|
117 |
|
@@ -120,7 +119,7 @@ class LibriVoxIndonesia(datasets.GeneratorBasedBuilder):
|
|
120 |
name=split_type[split],
|
121 |
gen_kwargs={
|
122 |
"local_extracted_archive": local_extracted_archive[split],
|
123 |
-
"audio_files": dl_manager.iter_archive(
|
124 |
"metadata_path": dl_manager.download_and_extract(metadata_path[split]),
|
125 |
"path_to_clips": path_to_clips,
|
126 |
},
|
|
|
30 |
|
31 |
_LICENSE = "https://creativecommons.org/publicdomain/zero/1.0/"
|
32 |
|
33 |
+
_DATA_URL = "https://huggingface.co/datasets/indonesian-nlp/librivox-indonesia/resolve/main/data"
|
34 |
|
35 |
|
36 |
class LibriVoxIndonesiaConfig(datasets.BuilderConfig):
|
|
|
41 |
self.release_date = kwargs.pop("release_date", None)
|
42 |
self.num_clips = kwargs.pop("num_clips", None)
|
43 |
self.num_speakers = kwargs.pop("num_speakers", None)
|
|
|
44 |
self.total_hr = kwargs.pop("total_hr", None)
|
45 |
self.size_bytes = kwargs.pop("size_bytes", None)
|
46 |
self.size_human = size_str(self.size_bytes)
|
47 |
description = (
|
48 |
f"LibriVox-Indonesia speech to text dataset in {self.language} released on {self.release_date}. "
|
49 |
+
f"The dataset comprises {self.total_hr} hours of transcribed speech data"
|
50 |
)
|
51 |
super(LibriVoxIndonesiaConfig, self).__init__(
|
52 |
name=name,
|
|
|
104 |
def _split_generators(self, dl_manager):
|
105 |
"""Returns SplitGenerators."""
|
106 |
dl_manager.download_config.ignore_url_params = True
|
107 |
+
audio_path = {}
|
108 |
local_extracted_archive = {}
|
109 |
metadata_path = {}
|
110 |
split_type = {"train": datasets.Split.TRAIN, "test": datasets.Split.TEST}
|
111 |
for split in split_type:
|
112 |
+
audio_path[split] = dl_manager.download(f"{_DATA_URL}/audio_{split}.tgz")
|
113 |
+
local_extracted_archive[split] = dl_manager.extract(audio_path[split]) if not dl_manager.is_streaming else None
|
114 |
metadata_path[split] = dl_manager.download_and_extract(f"{_DATA_URL}/metadata_{split}.csv.gz")
|
115 |
path_to_clips = "librivox-indonesia"
|
116 |
|
|
|
119 |
name=split_type[split],
|
120 |
gen_kwargs={
|
121 |
"local_extracted_archive": local_extracted_archive[split],
|
122 |
+
"audio_files": dl_manager.iter_archive(audio_path[split]),
|
123 |
"metadata_path": dl_manager.download_and_extract(metadata_path[split]),
|
124 |
"path_to_clips": path_to_clips,
|
125 |
},
|