Bharat Ramanathan commited on
Commit
a809d84
1 Parent(s): 229a75e

add first version of the mile dataset

Browse files
Files changed (2) hide show
  1. README.md +161 -0
  2. mile_dataset.py +136 -0
README.md ADDED
@@ -0,0 +1,161 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ annotations_creators:
3
+ - expert-generated
4
+ language:
5
+ - ta
6
+ language_creators:
7
+ - expert-generated
8
+ license:
9
+ - cc-by-2.0
10
+ multilinguality:
11
+ - monolingual
12
+ pretty_name: IISc-MILE Tamil ASR Corpus
13
+ size_categories:
14
+ - 10K<n<100K
15
+ source_datasets:
16
+ - original
17
+ tags:
18
+ - Tamil ASR
19
+ - Speech Recognition
20
+ task_categories:
21
+ - automatic-speech-recognition
22
+ task_ids: []
23
+ ---
24
+
25
+ # Dataset Card for [Dataset Name]
26
+
27
+ ## Table of Contents
28
+ - [Table of Contents](#table-of-contents)
29
+ - [Dataset Description](#dataset-description)
30
+ - [Dataset Summary](#dataset-summary)
31
+ - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
32
+ - [Languages](#languages)
33
+ - [Dataset Structure](#dataset-structure)
34
+ - [Data Instances](#data-instances)
35
+ - [Data Fields](#data-fields)
36
+ - [Data Splits](#data-splits)
37
+ - [Dataset Creation](#dataset-creation)
38
+ - [Curation Rationale](#curation-rationale)
39
+ - [Source Data](#source-data)
40
+ - [Annotations](#annotations)
41
+ - [Personal and Sensitive Information](#personal-and-sensitive-information)
42
+ - [Considerations for Using the Data](#considerations-for-using-the-data)
43
+ - [Social Impact of Dataset](#social-impact-of-dataset)
44
+ - [Discussion of Biases](#discussion-of-biases)
45
+ - [Other Known Limitations](#other-known-limitations)
46
+ - [Additional Information](#additional-information)
47
+ - [Dataset Curators](#dataset-curators)
48
+ - [Licensing Information](#licensing-information)
49
+ - [Citation Information](#citation-information)
50
+ - [Contributions](#contributions)
51
+
52
+ ## Dataset Description
53
+
54
+ - **Homepage:** https://www.openslr.org/127/
55
+ - **Repository:** https://github.com/MILE-IISc
56
+ - **Paper:** https://arxiv.org/abs/2207.13331
57
+ - **Leaderboard:**
58
+ - **Point of Contact:**
59
+
60
+ ### Dataset Summary
61
+
62
+ Tamil transcribed speech corpus for ASR
63
+
64
+ ### Supported Tasks and Leaderboards
65
+
66
+ [More Information Needed]
67
+
68
+ ### Languages
69
+
70
+ - Tamil
71
+
72
+ ## Dataset Structure
73
+
74
+ ### Data Instances
75
+
76
+ [More Information Needed]
77
+
78
+ ### Data Fields
79
+
80
+ [More Information Needed]
81
+
82
+ ### Data Splits
83
+
84
+ [More Information Needed]
85
+
86
+ ## Dataset Creation
87
+
88
+ ### Curation Rationale
89
+
90
+ [More Information Needed]
91
+
92
+ ### Source Data
93
+
94
+ #### Initial Data Collection and Normalization
95
+
96
+ [More Information Needed]
97
+
98
+ #### Who are the source language producers?
99
+
100
+ [More Information Needed]
101
+
102
+ ### Annotations
103
+
104
+ #### Annotation process
105
+
106
+ [More Information Needed]
107
+
108
+ #### Who are the annotators?
109
+
110
+ [More Information Needed]
111
+
112
+ ### Personal and Sensitive Information
113
+
114
+ [More Information Needed]
115
+
116
+ ## Considerations for Using the Data
117
+
118
+ ### Social Impact of Dataset
119
+
120
+ [More Information Needed]
121
+
122
+ ### Discussion of Biases
123
+
124
+ [More Information Needed]
125
+
126
+ ### Other Known Limitations
127
+
128
+ [More Information Needed]
129
+
130
+ ## Additional Information
131
+
132
+ ### Dataset Curators
133
+
134
+ [More Information Needed]
135
+
136
+ ### Licensing Information
137
+
138
+ Attribution 2.0 Generic (CC BY 2.0)
139
+
140
+ ### Citation Information
141
+ @misc{mile_1,
142
+ doi = {10.48550/ARXIV.2207.13331},
143
+ url = {https://arxiv.org/abs/2207.13331},
144
+ author = {A, Madhavaraj and Pilar, Bharathi and G, Ramakrishnan A},
145
+ title = {Subword Dictionary Learning and Segmentation Techniques for Automatic Speech Recognition in Tamil and Kannada},
146
+ publisher = {arXiv},
147
+ year = {2022},
148
+ }
149
+
150
+ @misc{mile_2,
151
+ doi = {10.48550/ARXIV.2207.13333},
152
+ url = {https://arxiv.org/abs/2207.13333},
153
+ author = {A, Madhavaraj and Pilar, Bharathi and G, Ramakrishnan A},
154
+ title = {Knowledge-driven Subword Grammar Modeling for Automatic Speech Recognition in Tamil and Kannada},
155
+ publisher = {arXiv},
156
+ year = {2022},
157
+ }
158
+
159
+ ### Contributions
160
+
161
+ Thanks to [@parambharat](https://github.com/parambharat) for adding this dataset.
mile_dataset.py ADDED
@@ -0,0 +1,136 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ """IISc-MILE Tamil ASR Corpus contains transcribed speech corpus for training ASR systems for Tamil language. It contains ~150 hours of read speech data collected from 531 speakers in a noise-free recording environment with high quality USB microphones. """
16
+
17
+
18
+ import json
19
+ import os
20
+
21
+ import datasets
22
+
23
+ _CITATION = """\
24
+ @misc{mile_1,
25
+ doi = {10.48550/ARXIV.2207.13331},
26
+ url = {https://arxiv.org/abs/2207.13331},
27
+ author = {A, Madhavaraj and Pilar, Bharathi and G, Ramakrishnan A},
28
+ title = {Subword Dictionary Learning and Segmentation Techniques for Automatic Speech Recognition in Tamil and Kannada},
29
+ publisher = {arXiv},
30
+ year = {2022},
31
+ }
32
+
33
+ @misc{mile_2,
34
+ doi = {10.48550/ARXIV.2207.13333},
35
+ url = {https://arxiv.org/abs/2207.13333},
36
+ author = {A, Madhavaraj and Pilar, Bharathi and G, Ramakrishnan A},
37
+ title = {Knowledge-driven Subword Grammar Modeling for Automatic Speech Recognition in Tamil and Kannada},
38
+ publisher = {arXiv},
39
+ year = {2022},
40
+ }
41
+ """
42
+
43
+ _DESCRIPTION = """\
44
+ IISc-MILE Tamil ASR Corpus contains transcribed speech corpus for training ASR systems for Tamil language. It contains ~150 hours of read speech data collected from 531 speakers in a noise-free recording environment with high quality USB microphones.
45
+ """
46
+
47
+ _HOMEPAGE = "https://www.openslr.org/127/"
48
+
49
+ _LICENSE = "Attribution 2.0 Generic (CC BY 2.0)"
50
+
51
+
52
+ _METADATA_URLS = {
53
+ "train": "data/train.jsonl",
54
+ "test": "data/test.jsonl"
55
+ }
56
+ _URLS = {
57
+ "train": "data/train.tar.gz",
58
+ "test": "data/test.tar.gz",
59
+
60
+ }
61
+
62
+ class MileDataset(datasets.GeneratorBasedBuilder):
63
+ """IISc-MILE Tamil ASR Corpus contains transcribed speech corpus for training ASR systems for Tamil language."""
64
+
65
+ VERSION = datasets.Version("1.1.0")
66
+ def _info(self):
67
+ features = datasets.Features(
68
+ {
69
+ "audio": datasets.Audio(sampling_rate=16_000),
70
+ "file_name": datasets.Value("string"),
71
+ "sentence": datasets.Value("string"),
72
+ }
73
+ )
74
+ return datasets.DatasetInfo(
75
+ description=_DESCRIPTION,
76
+ features=features,
77
+ supervised_keys=("sentence", "label"),
78
+ homepage=_HOMEPAGE,
79
+ license=_LICENSE,
80
+ citation=_CITATION,
81
+ )
82
+
83
+ def _split_generators(self, dl_manager):
84
+ metadata_paths = dl_manager.download(_METADATA_URLS)
85
+ train_archive = dl_manager.download(_URLS["train"])
86
+ test_archive = dl_manager.download(_URLS["test"])
87
+ local_extracted_train_archive = dl_manager.extract(train_archive) if not dl_manager.is_streaming else None
88
+ local_extracted_test_archive = dl_manager.extract(test_archive) if not dl_manager.is_streaming else None
89
+ test_archive = dl_manager.download(_URLS["test"])
90
+ train_dir = "train"
91
+ test_dir = "test"
92
+
93
+ return [
94
+ datasets.SplitGenerator(
95
+ name=datasets.Split.TRAIN,
96
+ gen_kwargs={
97
+ "metadata_path": metadata_paths["train"],
98
+ "local_extracted_archive": local_extracted_train_archive,
99
+ "path_to_clips": train_dir + "/mp3",
100
+ "audio_files": dl_manager.iter_archive(train_archive),
101
+ },
102
+ ),
103
+ datasets.SplitGenerator(
104
+ name=datasets.Split.TEST,
105
+ gen_kwargs={
106
+ "metadata_path": metadata_paths["test"],
107
+ "local_extracted_archive": local_extracted_test_archive,
108
+ "path_to_clips": test_dir + "/mp3",
109
+ "audio_files": dl_manager.iter_archive(test_archive),
110
+ },
111
+ ),
112
+
113
+ ]
114
+
115
+ def _generate_examples(self, metadata_path, local_extracted_archive, path_to_clips, audio_files):
116
+ """Yields examples as (key, example) tuples."""
117
+ examples = {}
118
+ with open(metadata_path, encoding="utf-8") as f:
119
+ for key, row in enumerate(f):
120
+ data = json.loads(row)
121
+ examples[data["file_name"]] = data
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
+ result = examples[path]
129
+ path = os.path.join(local_extracted_archive, path) if local_extracted_archive else path
130
+ result["audio"] = {"path": path, "bytes": f.read()}
131
+ result["file_name"] = path
132
+ yield id_, result
133
+ id_ += 1
134
+ elif inside_clips_dir:
135
+ break
136
+