test new folder structure
Browse files- .gitattributes +1 -0
- data/2015-other.tar → audio/clean/test/clean_test.tar +2 -2
- audio/clean/train/clean_train.tar +3 -0
- data/clean/test/clean_test.tsv +3 -0
- data/clean/train/clean_train.tsv +3 -0
- data/transcript.tsv +0 -2
- my_dataset.py +0 -132
- nos-parlaspeech.py +94 -0
.gitattributes
CHANGED
@@ -54,3 +54,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
54 |
*.jpeg filter=lfs diff=lfs merge=lfs -text
|
55 |
*.webp filter=lfs diff=lfs merge=lfs -text
|
56 |
2015-other.tar filter=lfs diff=lfs merge=lfs -text
|
|
|
|
54 |
*.jpeg filter=lfs diff=lfs merge=lfs -text
|
55 |
*.webp filter=lfs diff=lfs merge=lfs -text
|
56 |
2015-other.tar filter=lfs diff=lfs merge=lfs -text
|
57 |
+
*.tsv filter=lfs diff=lfs merge=lfs -text
|
data/2015-other.tar → audio/clean/test/clean_test.tar
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8ccbb447d99bb45a93a4b3aef70379bffaf88d9dd36248353c656883449556c9
|
3 |
+
size 2715852800
|
audio/clean/train/clean_train.tar
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1c5d6ff4daaee62df0f071b671f7ce02c8c7cc8a5039ced70144ffc477e687bc
|
3 |
+
size 7456389120
|
data/clean/test/clean_test.tsv
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:775e5335d84cfadda12e4e91c0c23d65d375f1a77b706ef93dea52e20d7d44cb
|
3 |
+
size 1856141
|
data/clean/train/clean_train.tsv
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:aeaac895cf84c48497176d49aac9b7f8d44df1389222274235e21bd1ce89df82
|
3 |
+
size 7898749
|
data/transcript.tsv
DELETED
@@ -1,2 +0,0 @@
|
|
1 |
-
path sentence
|
2 |
-
DSPG_137_23122015_9873.69_9888.03.wav THIS IS A SAMPLE SENTENCE
|
|
|
|
|
|
my_dataset.py
DELETED
@@ -1,132 +0,0 @@
|
|
1 |
-
import LANGUAGES as LANGUAGES
|
2 |
-
import STATS as STATS
|
3 |
-
import datasets as datasets
|
4 |
-
from datasets.utils.py_utils import size_str
|
5 |
-
|
6 |
-
_HOMEPAGE = "homepage-info"
|
7 |
-
_CITATION = "citation-info"
|
8 |
-
_LICENSE = "license-info"
|
9 |
-
_DESCRIPTION = "description-info"
|
10 |
-
|
11 |
-
_PROMPTS_URLS = "....."
|
12 |
-
_DATA_URL = "...."
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
"""Configuration class, allows to have multiple configurations if needed"""
|
17 |
-
class ParlaSpeechDatasetConfig(datasets.BuilderConfig):
|
18 |
-
"""BuilderConfig for ParlaSpeech"""
|
19 |
-
|
20 |
-
def __init__(self, name, version, **kwargs):
|
21 |
-
self.language = kwargs.pop("language", None)
|
22 |
-
self.release_date = kwargs.pop("release_date", None)
|
23 |
-
self.num_clips = kwargs.pop("num_clips", None)
|
24 |
-
self.num_speakers = kwargs.pop("num_speakers", None)
|
25 |
-
self.validated_hr = kwargs.pop("validated_hr", None)
|
26 |
-
self.total_hr = kwargs.pop("total_hr", None)
|
27 |
-
self.size_bytes = kwargs.pop("size_bytes", None)
|
28 |
-
self.size_human = size_str(self.size_bytes)
|
29 |
-
description = ( ##Update Description in the final version
|
30 |
-
f"ParlaSpeech is a dataset in {self.language} released on {self.release_date}. "
|
31 |
-
)
|
32 |
-
super(ParlaSpeechDatasetConfig, self).__init__(
|
33 |
-
name=name,
|
34 |
-
version=datasets.Version(version),
|
35 |
-
description=description,
|
36 |
-
**kwargs,
|
37 |
-
)
|
38 |
-
|
39 |
-
|
40 |
-
class ParlaSpeechDataset(datasets.GeneratroBasedBuilder):
|
41 |
-
|
42 |
-
""""
|
43 |
-
### NO TENGO CLARO SI HACE FALTA ESTO ###
|
44 |
-
DEFAULT_CONFIG_NAME = "all"
|
45 |
-
|
46 |
-
BUILDER_CONFIGS = [
|
47 |
-
ParlaSpeechDatasetConfig(
|
48 |
-
name=lang,
|
49 |
-
version=STATS["version"],
|
50 |
-
language=LANGUAGES[lang],
|
51 |
-
release_date=STATS["date"],
|
52 |
-
num_clips=lang_stats["clips"],
|
53 |
-
num_speakers=lang_stats["users"],
|
54 |
-
total_hr=float(lang_stats["totalHrs"]) if lang_stats["totalHrs"] else None,
|
55 |
-
size_bytes=int(lang_stats["size"]) if lang_stats["size"] else None,
|
56 |
-
)
|
57 |
-
for lang, lang_stats in STATS["locales"].items()
|
58 |
-
]
|
59 |
-
"""
|
60 |
-
|
61 |
-
""" When the dataset is loaded and .info is called, the info defined here is displayed."""
|
62 |
-
def _info(self):
|
63 |
-
return datasets.DatasetInfo(
|
64 |
-
description=_DESCRIPTION,
|
65 |
-
features=datasets.Features(
|
66 |
-
{
|
67 |
-
#"speaker_id": datasets.Value("string"),
|
68 |
-
#"path": datasets.Value("string"),
|
69 |
-
"path": datasets.Audio(sampling_rate=16_000),
|
70 |
-
"sentence": datasets.Value("string"),
|
71 |
-
}
|
72 |
-
),
|
73 |
-
supervised_keys=None,
|
74 |
-
homepage=_HOMEPAGE,
|
75 |
-
license=_LICENSE,
|
76 |
-
citation=_CITATION,
|
77 |
-
version = self.config.version,
|
78 |
-
)
|
79 |
-
|
80 |
-
" Used to organize the audio files and sentence prompts in each split, once downloaded the dataset."
|
81 |
-
def _split_generators(self, dl_manager):
|
82 |
-
"""Returns SplitGenerators"""
|
83 |
-
prompts_paths = dl_manager.download(_PROMPTS_URLS)
|
84 |
-
archive = dl_manager.download(_DATA_URL)
|
85 |
-
## local_extracted_archives = dl_manager.extract(archive)
|
86 |
-
train_dir = "vivos/train"
|
87 |
-
test_dir = "vivos/test"
|
88 |
-
|
89 |
-
return [
|
90 |
-
datasets.SplitGenerator(
|
91 |
-
name=datasets.Split.TRAIN,
|
92 |
-
gen_kwargs={
|
93 |
-
"prompts_path": prompts_paths["train"],
|
94 |
-
"path_to_clips": train_dir + "/waves",
|
95 |
-
"audio_files": dl_manager.iter_archive(archive),
|
96 |
-
},
|
97 |
-
),
|
98 |
-
datasets.SplitGenerator(
|
99 |
-
name=datasets.Split.TEST,
|
100 |
-
gen_kwargs={
|
101 |
-
"prompts_path": prompts_paths["test"],
|
102 |
-
"path_to_clips": test_dir + "/waves",
|
103 |
-
"audio_files": dl_manager.iter_archive(archive),
|
104 |
-
},
|
105 |
-
),
|
106 |
-
]
|
107 |
-
|
108 |
-
def _generate_examples(self, prompts_path, path_to_clips, audio_files):
|
109 |
-
"""Yields examples as (key, example) tuples."""
|
110 |
-
examples = {}
|
111 |
-
with open(prompts_path, encoding="utf-8") as f: ##prompts_path -> transcript.tsv
|
112 |
-
for row in f:
|
113 |
-
data = row.strip().split(" ", 1)
|
114 |
-
#speaker_id = data[0].split("_")[0]
|
115 |
-
#audio_path = "/".join([path_to_clips, speaker_id, data[0] + ".wav"])
|
116 |
-
audio_path = "/".join([path_to_clips, "DSPG_137_23122015_9873.69_9888.03.wav"])
|
117 |
-
examples[audio_path] = {
|
118 |
-
#"speaker_id": speaker_id,
|
119 |
-
"path": audio_path,
|
120 |
-
"sentence": data[1],
|
121 |
-
}
|
122 |
-
inside_clips_dir = False
|
123 |
-
id_ = 0
|
124 |
-
for path, f in audio_files:
|
125 |
-
if path.startswith(path_to_clips):
|
126 |
-
inside_clips_dir = True
|
127 |
-
if path in examples:
|
128 |
-
audio = {"path": path, "bytes": f.read()}
|
129 |
-
yield id_, {**examples[path], "audio": audio}
|
130 |
-
id_ += 1
|
131 |
-
elif inside_clips_dir:
|
132 |
-
break
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
nos-parlaspeech.py
ADDED
@@ -0,0 +1,94 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# coding=utf-8
|
2 |
+
|
3 |
+
import pandas as pd
|
4 |
+
|
5 |
+
import datasets
|
6 |
+
from datasets.tasks import AutomaticSpeechRecognition
|
7 |
+
|
8 |
+
|
9 |
+
_CITATION = """\
|
10 |
+
|
11 |
+
"""
|
12 |
+
|
13 |
+
_DESCRIPTION = """\
|
14 |
+
|
15 |
+
"""
|
16 |
+
|
17 |
+
_HOMEPAGE = "https://zenodo.org/record/5541827"
|
18 |
+
|
19 |
+
_LICENSE = "Creative Commons Attribution 4.0 International"
|
20 |
+
|
21 |
+
_INDEX_REPO = "https://huggingface.co/datasets/proxectonos/Nos_Parlaspeech-GL/tree/main/"
|
22 |
+
#_INDEX_REPO = "https://huggingface.co/datasets/proxectonos/Nos_Parlaspeech-GL/resolve/main/"
|
23 |
+
|
24 |
+
_URLS = {
|
25 |
+
"index": _INDEX_REPO + "data/{config}/{split}/{config}_{split}.tsv",
|
26 |
+
"audio": "audio/{config}/{split}/{config}_{split}.tar",
|
27 |
+
#"audio": "audio/{config}/{split}/{config}_{split}.tar?download=1",
|
28 |
+
}
|
29 |
+
_SPLITS = {datasets.Split.TRAIN: "train", datasets.Split.VALIDATION: "dev", datasets.Split.TEST: "test"}
|
30 |
+
|
31 |
+
|
32 |
+
class ParlaSpeech(datasets.GeneratorBasedBuilder):
|
33 |
+
"""Nos-ParlaSpeech."""
|
34 |
+
|
35 |
+
VERSION = datasets.Version("1.0")
|
36 |
+
|
37 |
+
BUILDER_CONFIGS = [
|
38 |
+
datasets.BuilderConfig(name="clean", version=VERSION, description="XX hours of clean quality segments."),
|
39 |
+
datasets.BuilderConfig(name="other", version=VERSION, description="XX hours of other quality segments."),
|
40 |
+
]
|
41 |
+
|
42 |
+
def _info(self):
|
43 |
+
return datasets.DatasetInfo(
|
44 |
+
description=_DESCRIPTION,
|
45 |
+
features=datasets.Features(
|
46 |
+
{
|
47 |
+
"path": datasets.Value("string"), #path to wav file
|
48 |
+
"audio": datasets.features.Audio(),
|
49 |
+
"speaker_id": datasets.Value("int64"),
|
50 |
+
"sentence": datasets.Value("string"),
|
51 |
+
"gender": datasets.ClassLabel(names=["F", "M"]),
|
52 |
+
"duration": datasets.Value("float64"),
|
53 |
+
}
|
54 |
+
),
|
55 |
+
supervised_keys=None,
|
56 |
+
homepage=_HOMEPAGE,
|
57 |
+
license=_LICENSE,
|
58 |
+
citation=_CITATION,
|
59 |
+
task_templates=[
|
60 |
+
AutomaticSpeechRecognition(transcription_column="sentence")
|
61 |
+
],
|
62 |
+
)
|
63 |
+
|
64 |
+
def _split_generators(self, dl_manager):
|
65 |
+
urls = {
|
66 |
+
split: {key: url.format(config=self.config.name, split=_SPLITS[split]) for key, url in _URLS.items()}
|
67 |
+
for split in _SPLITS
|
68 |
+
}
|
69 |
+
dl_dir = dl_manager.download(urls)
|
70 |
+
return [
|
71 |
+
datasets.SplitGenerator(
|
72 |
+
name=split,
|
73 |
+
gen_kwargs={
|
74 |
+
"index_path": dl_dir[split]["index"],
|
75 |
+
"audio_files": dl_manager.iter_archive(dl_dir[split]["audio"]),
|
76 |
+
},
|
77 |
+
)
|
78 |
+
for split in _SPLITS
|
79 |
+
]
|
80 |
+
|
81 |
+
def _generate_examples(self, index_path, audio_files):
|
82 |
+
with open(index_path, encoding="utf-8") as index_file:
|
83 |
+
index = pd.read_csv(index_file, delimiter="\t", index_col="path").to_dict(orient="index")
|
84 |
+
# clean: 83568 = 79269 + 2155 + 2144 ; other: 146669 = 142813 + 1957 + 1899
|
85 |
+
for key, (path, file) in enumerate(audio_files):
|
86 |
+
if path.endswith(".wav"):
|
87 |
+
data = index.pop(path)
|
88 |
+
audio = {"path": path, "bytes": file.read()}
|
89 |
+
yield key, {"path": path, "audio": audio, **data}
|
90 |
+
else:
|
91 |
+
path = path + ".wav"
|
92 |
+
data = index.pop(path)
|
93 |
+
audio = {"path": path, "bytes": file.read()}
|
94 |
+
yield key, {"path": path, "audio": audio, **data}
|