lhoestq HF staff ylacombe HF staff commited on
Commit
2e83e61
1 Parent(s): 92be018

mls_without_script (#15)

Browse files

- Upload dataset (56cb16aa5347685f0059abd744f098c2e0716eb1)
- Upload dataset (b2688928afd3a23afaca076626ee58346feef469)
- Upload dataset (c82617d933b9de71a3ceebce12cc2d8da66bc319)
- Upload dataset (8ac09f12ed9cc3a6676c13be759a02de56af3eea)
- Upload dataset (part 00000-of-00002) (124f036150ad37664355f25c1cb79988b674f199)
- Upload dataset (part 00001-of-00002) (12bf0eb4b1e903aafdb0a4c39f2353872e07d8f6)
- Upload dataset (0d33592b032b0c6d9c8fc2530bc7f48147f7b9d9)
- Upload dataset (part 00000-of-00002) (0a4a1bac8fbb68632722ced68b81645c1faec46b)
- Upload dataset (part 00001-of-00002) (c22a1da6206b7a069e9d8d1e69ac423b771e560d)
- Update README.md (326dcbdd5fd51afc48fc11d276991c2b404317c5)
- Create create_dataset.py (8a439ef1cea73b319377c4f688a691359c91017f)
- Delete multilingual_librispeech.py (314c28d8ff4b9546bd8e1c956731ece5e6984158)


Co-authored-by: Yoach Lacombe <[email protected]>

This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. README.md +416 -14
  2. create_dataset.py +106 -0
  3. dutch/1_hours-00000-of-00001.parquet +3 -0
  4. dutch/9_hours-00000-of-00001.parquet +3 -0
  5. dutch/dev-00000-of-00001.parquet +3 -0
  6. dutch/test-00000-of-00001.parquet +3 -0
  7. dutch/train-00000-of-00048.parquet +3 -0
  8. dutch/train-00001-of-00048.parquet +3 -0
  9. dutch/train-00002-of-00048.parquet +3 -0
  10. dutch/train-00003-of-00048.parquet +3 -0
  11. dutch/train-00004-of-00048.parquet +3 -0
  12. dutch/train-00005-of-00048.parquet +3 -0
  13. dutch/train-00006-of-00048.parquet +3 -0
  14. dutch/train-00007-of-00048.parquet +3 -0
  15. dutch/train-00008-of-00048.parquet +3 -0
  16. dutch/train-00009-of-00048.parquet +3 -0
  17. dutch/train-00010-of-00048.parquet +3 -0
  18. dutch/train-00011-of-00048.parquet +3 -0
  19. dutch/train-00012-of-00048.parquet +3 -0
  20. dutch/train-00013-of-00048.parquet +3 -0
  21. dutch/train-00014-of-00048.parquet +3 -0
  22. dutch/train-00015-of-00048.parquet +3 -0
  23. dutch/train-00016-of-00048.parquet +3 -0
  24. dutch/train-00017-of-00048.parquet +3 -0
  25. dutch/train-00018-of-00048.parquet +3 -0
  26. dutch/train-00019-of-00048.parquet +3 -0
  27. dutch/train-00020-of-00048.parquet +3 -0
  28. dutch/train-00021-of-00048.parquet +3 -0
  29. dutch/train-00022-of-00048.parquet +3 -0
  30. dutch/train-00023-of-00048.parquet +3 -0
  31. dutch/train-00024-of-00048.parquet +3 -0
  32. dutch/train-00025-of-00048.parquet +3 -0
  33. dutch/train-00026-of-00048.parquet +3 -0
  34. dutch/train-00027-of-00048.parquet +3 -0
  35. dutch/train-00028-of-00048.parquet +3 -0
  36. dutch/train-00029-of-00048.parquet +3 -0
  37. dutch/train-00030-of-00048.parquet +3 -0
  38. dutch/train-00031-of-00048.parquet +3 -0
  39. dutch/train-00032-of-00048.parquet +3 -0
  40. dutch/train-00033-of-00048.parquet +3 -0
  41. dutch/train-00034-of-00048.parquet +3 -0
  42. dutch/train-00035-of-00048.parquet +3 -0
  43. dutch/train-00036-of-00048.parquet +3 -0
  44. dutch/train-00037-of-00048.parquet +3 -0
  45. dutch/train-00038-of-00048.parquet +3 -0
  46. dutch/train-00039-of-00048.parquet +3 -0
  47. dutch/train-00040-of-00048.parquet +3 -0
  48. dutch/train-00041-of-00048.parquet +3 -0
  49. dutch/train-00042-of-00048.parquet +3 -0
  50. dutch/train-00043-of-00048.parquet +3 -0
README.md CHANGED
@@ -1,5 +1,4 @@
1
  ---
2
- pretty_name: MultiLingual LibriSpeech
3
  annotations_creators:
4
  - expert-generated
5
  language_creators:
@@ -13,17 +12,387 @@ language:
13
  - es
14
  - pt
15
  - pl
 
16
  license:
17
  - cc-by-4.0
18
  multilinguality:
19
  - multilingual
20
- paperswithcode_id: multilingual-librispeech
21
  size_categories:
22
  - 100K<n<1M
23
  source_datasets:
24
  - original
25
  task_categories:
26
  - automatic-speech-recognition
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
27
  ---
28
 
29
  # Dataset Card for MultiLingual LibriSpeech
@@ -66,11 +435,12 @@ This is a streamable version of the Multilingual LibriSpeech (MLS) dataset.
66
  The data archives were restructured from the original ones from [OpenSLR](http://www.openslr.org/94) to make it easier to stream.
67
 
68
  MLS dataset is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of
69
- 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish.
70
 
71
  ### Supported Tasks and Leaderboards
72
 
73
  - `automatic-speech-recognition`, `speaker-identification`: The dataset can be used to train a model for Automatic Speech Recognition (ASR). The model is presented with an audio file and asked to transcribe the audio file to written text. The most common evaluation metric is the word error rate (WER). The task has an active leaderboard which can be found at https://paperswithcode.com/dataset/multilingual-librispeech and ranks models based on their WER.
 
74
 
75
  ### Languages
76
 
@@ -83,16 +453,13 @@ The `datasets` library allows you to load and pre-process your dataset in pure P
83
  For example, to download the German config, simply specify the corresponding language config name (i.e., "german" for German):
84
  ```python
85
  from datasets import load_dataset
86
-
87
  mls = load_dataset("facebook/multilingual_librispeech", "german", split="train")
88
  ```
89
 
90
  Using the datasets library, you can also stream the dataset on-the-fly by adding a `streaming=True` argument to the `load_dataset` function call. Loading a dataset in streaming mode loads individual samples of the dataset at a time, rather than downloading the entire dataset to disk.
91
  ```python
92
  from datasets import load_dataset
93
-
94
  mls = load_dataset("facebook/multilingual_librispeech", "german", split="train", streaming=True)
95
-
96
  print(next(iter(mls)))
97
  ```
98
 
@@ -103,7 +470,6 @@ Local:
103
  ```python
104
  from datasets import load_dataset
105
  from torch.utils.data.sampler import BatchSampler, RandomSampler
106
-
107
  mls = load_dataset("facebook/multilingual_librispeech", "german", split="train")
108
  batch_sampler = BatchSampler(RandomSampler(mls), batch_size=32, drop_last=False)
109
  dataloader = DataLoader(mls, batch_sampler=batch_sampler)
@@ -114,7 +480,6 @@ Streaming:
114
  ```python
115
  from datasets import load_dataset
116
  from torch.utils.data import DataLoader
117
-
118
  mls = load_dataset("facebook/multilingual_librispeech", "german", split="train", streaming=True)
119
  dataloader = DataLoader(mls, batch_size=32)
120
  ```
@@ -155,12 +520,11 @@ A typical data point comprises the path to the audio file, usually called `file`
155
  - id: unique id of the data sample.
156
 
157
  - speaker_id: unique id of the speaker. The same speaker id can be found for multiple data samples.
158
-
159
  - chapter_id: id of the audiobook chapter which includes the transcription.
160
 
161
  ### Data Splits
162
 
163
- | | Train | Train.9h | Train.1h | Dev | Test |
164
  | ----- | ------ | ----- | ---- | ---- | ---- |
165
  | german | 469942 | 2194 | 241 | 3469 | 3394 |
166
  | dutch | 374287 | 2153 | 234 | 3095 | 3075 |
@@ -170,8 +534,6 @@ A typical data point comprises the path to the audio file, usually called `file`
170
  | portuguese | 37533 | 2116 | 236 | 826 | 871 |
171
  | polish | 25043 | 2173 | 238 | 512 | 520 |
172
 
173
-
174
-
175
  ## Dataset Creation
176
 
177
  ### Curation Rationale
@@ -238,7 +600,47 @@ Public Domain, Creative Commons Attribution 4.0 International Public License ([C
238
  }
239
  ```
240
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
241
  ### Contributions
242
 
243
- Thanks to [@patrickvonplaten](https://github.com/patrickvonplaten)
244
- and [@polinaeterna](https://github.com/polinaeterna) for adding this dataset.
 
1
  ---
 
2
  annotations_creators:
3
  - expert-generated
4
  language_creators:
 
12
  - es
13
  - pt
14
  - pl
15
+ - en
16
  license:
17
  - cc-by-4.0
18
  multilinguality:
19
  - multilingual
 
20
  size_categories:
21
  - 100K<n<1M
22
  source_datasets:
23
  - original
24
  task_categories:
25
  - automatic-speech-recognition
26
+ - text-to-speech
27
+ - text-to-audio
28
+ paperswithcode_id: multilingual-librispeech
29
+ pretty_name: MultiLingual LibriSpeech
30
+ dataset_info:
31
+ - config_name: dutch
32
+ features:
33
+ - name: audio
34
+ dtype: audio
35
+ - name: original_path
36
+ dtype: string
37
+ - name: begin_time
38
+ dtype: float64
39
+ - name: end_time
40
+ dtype: float64
41
+ - name: transcript
42
+ dtype: string
43
+ - name: audio_duration
44
+ dtype: float64
45
+ - name: speaker_id
46
+ dtype: string
47
+ - name: chapter_id
48
+ dtype: string
49
+ - name: file
50
+ dtype: string
51
+ - name: id
52
+ dtype: string
53
+ splits:
54
+ - name: dev
55
+ num_bytes: 199959986
56
+ num_examples: 3095
57
+ - name: test
58
+ num_bytes: 199298575
59
+ num_examples: 3075
60
+ - name: train
61
+ num_bytes: 23931679031
62
+ num_examples: 374287
63
+ - name: 9_hours
64
+ num_bytes: 139884664.668
65
+ num_examples: 2153
66
+ - name: 1_hours
67
+ num_bytes: 15462181
68
+ num_examples: 234
69
+ download_size: 24376256629
70
+ dataset_size: 24486284437.668
71
+ - config_name: french
72
+ features:
73
+ - name: audio
74
+ dtype: audio
75
+ - name: original_path
76
+ dtype: string
77
+ - name: begin_time
78
+ dtype: float64
79
+ - name: end_time
80
+ dtype: float64
81
+ - name: transcript
82
+ dtype: string
83
+ - name: audio_duration
84
+ dtype: float64
85
+ - name: speaker_id
86
+ dtype: string
87
+ - name: chapter_id
88
+ dtype: string
89
+ - name: file
90
+ dtype: string
91
+ - name: id
92
+ dtype: string
93
+ splits:
94
+ - name: dev
95
+ num_bytes: 157923970.696
96
+ num_examples: 2416
97
+ - name: test
98
+ num_bytes: 158352158.582
99
+ num_examples: 2426
100
+ - name: train
101
+ num_bytes: 16984935842.04
102
+ num_examples: 258213
103
+ - name: 9_hours
104
+ num_bytes: 142796680.609
105
+ num_examples: 2167
106
+ - name: 1_hours
107
+ num_bytes: 15675831
108
+ num_examples: 241
109
+ download_size: 17381581776
110
+ dataset_size: 17459684482.927002
111
+ - config_name: german
112
+ features:
113
+ - name: audio
114
+ dtype: audio
115
+ - name: original_path
116
+ dtype: string
117
+ - name: begin_time
118
+ dtype: float64
119
+ - name: end_time
120
+ dtype: float64
121
+ - name: transcript
122
+ dtype: string
123
+ - name: audio_duration
124
+ dtype: float64
125
+ - name: speaker_id
126
+ dtype: string
127
+ - name: chapter_id
128
+ dtype: string
129
+ - name: file
130
+ dtype: string
131
+ - name: id
132
+ dtype: string
133
+ splits:
134
+ - name: dev
135
+ num_bytes: 224293581.302
136
+ num_examples: 3469
137
+ - name: test
138
+ num_bytes: 225756069.096
139
+ num_examples: 3394
140
+ - name: train
141
+ num_bytes: 31050881388
142
+ num_examples: 469942
143
+ - name: 9_hours
144
+ num_bytes: 142777983.118
145
+ num_examples: 2194
146
+ - name: 1_hours
147
+ num_bytes: 15714704
148
+ num_examples: 241
149
+ download_size: 31526161821
150
+ dataset_size: 31659423725.516
151
+ - config_name: italian
152
+ features:
153
+ - name: audio
154
+ dtype: audio
155
+ - name: original_path
156
+ dtype: string
157
+ - name: begin_time
158
+ dtype: float64
159
+ - name: end_time
160
+ dtype: float64
161
+ - name: transcript
162
+ dtype: string
163
+ - name: audio_duration
164
+ dtype: float64
165
+ - name: speaker_id
166
+ dtype: string
167
+ - name: chapter_id
168
+ dtype: string
169
+ - name: file
170
+ dtype: string
171
+ - name: id
172
+ dtype: string
173
+ splits:
174
+ - name: dev
175
+ num_bytes: 81607596.048
176
+ num_examples: 1248
177
+ - name: test
178
+ num_bytes: 83216752.046
179
+ num_examples: 1262
180
+ - name: train
181
+ num_bytes: 3896742625
182
+ num_examples: 59623
183
+ - name: 9_hours
184
+ num_bytes: 141671904.428
185
+ num_examples: 2173
186
+ - name: 1_hours
187
+ num_bytes: 15560398
188
+ num_examples: 240
189
+ download_size: 4200633596
190
+ dataset_size: 4218799275.522
191
+ - config_name: polish
192
+ features:
193
+ - name: audio
194
+ dtype: audio
195
+ - name: original_path
196
+ dtype: string
197
+ - name: begin_time
198
+ dtype: float64
199
+ - name: end_time
200
+ dtype: float64
201
+ - name: transcript
202
+ dtype: string
203
+ - name: audio_duration
204
+ dtype: float64
205
+ - name: speaker_id
206
+ dtype: string
207
+ - name: chapter_id
208
+ dtype: string
209
+ - name: file
210
+ dtype: string
211
+ - name: id
212
+ dtype: string
213
+ splits:
214
+ - name: dev
215
+ num_bytes: 32746725
216
+ num_examples: 512
217
+ - name: test
218
+ num_bytes: 33735044
219
+ num_examples: 520
220
+ - name: train
221
+ num_bytes: 1638889846
222
+ num_examples: 25043
223
+ - name: 9_hours
224
+ num_bytes: 142005461
225
+ num_examples: 2173
226
+ - name: 1_hours
227
+ num_bytes: 15681216
228
+ num_examples: 238
229
+ download_size: 1855342312
230
+ dataset_size: 1863058292
231
+ - config_name: portuguese
232
+ features:
233
+ - name: audio
234
+ dtype: audio
235
+ - name: original_path
236
+ dtype: string
237
+ - name: begin_time
238
+ dtype: float64
239
+ - name: end_time
240
+ dtype: float64
241
+ - name: transcript
242
+ dtype: string
243
+ - name: audio_duration
244
+ dtype: float64
245
+ - name: speaker_id
246
+ dtype: string
247
+ - name: chapter_id
248
+ dtype: string
249
+ - name: file
250
+ dtype: string
251
+ - name: id
252
+ dtype: string
253
+ splits:
254
+ - name: dev
255
+ num_bytes: 57533473
256
+ num_examples: 826
257
+ - name: test
258
+ num_bytes: 59141979
259
+ num_examples: 871
260
+ - name: train
261
+ num_bytes: 2518553713.946
262
+ num_examples: 37533
263
+ - name: 9_hours
264
+ num_bytes: 141641902.42
265
+ num_examples: 2116
266
+ - name: 1_hours
267
+ num_bytes: 15697139
268
+ num_examples: 236
269
+ download_size: 2780836500
270
+ dataset_size: 2792568207.366
271
+ - config_name: spanish
272
+ features:
273
+ - name: audio
274
+ dtype: audio
275
+ - name: original_path
276
+ dtype: string
277
+ - name: begin_time
278
+ dtype: float64
279
+ - name: end_time
280
+ dtype: float64
281
+ - name: transcript
282
+ dtype: string
283
+ - name: audio_duration
284
+ dtype: float64
285
+ - name: speaker_id
286
+ dtype: string
287
+ - name: chapter_id
288
+ dtype: string
289
+ - name: file
290
+ dtype: string
291
+ - name: id
292
+ dtype: string
293
+ splits:
294
+ - name: dev
295
+ num_bytes: 157804903.144
296
+ num_examples: 2408
297
+ - name: test
298
+ num_bytes: 158526899.32
299
+ num_examples: 2385
300
+ - name: train
301
+ num_bytes: 14562584188
302
+ num_examples: 220701
303
+ - name: 9_hours
304
+ num_bytes: 142473624.48
305
+ num_examples: 2110
306
+ - name: 1_hours
307
+ num_bytes: 15702048
308
+ num_examples: 233
309
+ download_size: 14971394533
310
+ dataset_size: 15037091662.944
311
+ configs:
312
+ - config_name: dutch
313
+ data_files:
314
+ - split: dev
315
+ path: dutch/dev-*
316
+ - split: test
317
+ path: dutch/test-*
318
+ - split: train
319
+ path: dutch/train-*
320
+ - split: 9_hours
321
+ path: dutch/9_hours-*
322
+ - split: 1_hours
323
+ path: dutch/1_hours-*
324
+ - config_name: french
325
+ data_files:
326
+ - split: dev
327
+ path: french/dev-*
328
+ - split: test
329
+ path: french/test-*
330
+ - split: train
331
+ path: french/train-*
332
+ - split: 9_hours
333
+ path: french/9_hours-*
334
+ - split: 1_hours
335
+ path: french/1_hours-*
336
+ - config_name: german
337
+ data_files:
338
+ - split: dev
339
+ path: german/dev-*
340
+ - split: test
341
+ path: german/test-*
342
+ - split: train
343
+ path: german/train-*
344
+ - split: 9_hours
345
+ path: german/9_hours-*
346
+ - split: 1_hours
347
+ path: german/1_hours-*
348
+ - config_name: italian
349
+ data_files:
350
+ - split: dev
351
+ path: italian/dev-*
352
+ - split: test
353
+ path: italian/test-*
354
+ - split: train
355
+ path: italian/train-*
356
+ - split: 9_hours
357
+ path: italian/9_hours-*
358
+ - split: 1_hours
359
+ path: italian/1_hours-*
360
+ - config_name: polish
361
+ data_files:
362
+ - split: dev
363
+ path: polish/dev-*
364
+ - split: test
365
+ path: polish/test-*
366
+ - split: train
367
+ path: polish/train-*
368
+ - split: 9_hours
369
+ path: polish/9_hours-*
370
+ - split: 1_hours
371
+ path: polish/1_hours-*
372
+ - config_name: portuguese
373
+ data_files:
374
+ - split: dev
375
+ path: portuguese/dev-*
376
+ - split: test
377
+ path: portuguese/test-*
378
+ - split: train
379
+ path: portuguese/train-*
380
+ - split: 9_hours
381
+ path: portuguese/9_hours-*
382
+ - split: 1_hours
383
+ path: portuguese/1_hours-*
384
+ - config_name: spanish
385
+ data_files:
386
+ - split: dev
387
+ path: spanish/dev-*
388
+ - split: test
389
+ path: spanish/test-*
390
+ - split: train
391
+ path: spanish/train-*
392
+ - split: 9_hours
393
+ path: spanish/9_hours-*
394
+ - split: 1_hours
395
+ path: spanish/1_hours-*
396
  ---
397
 
398
  # Dataset Card for MultiLingual LibriSpeech
 
435
  The data archives were restructured from the original ones from [OpenSLR](http://www.openslr.org/94) to make it easier to stream.
436
 
437
  MLS dataset is a large multilingual corpus suitable for speech research. The dataset is derived from read audiobooks from LibriVox and consists of
438
+ 8 languages - English, German, Dutch, Spanish, French, Italian, Portuguese, Polish. It includes about 44.5K hours of English and a total of about 6K hours for other languages.
439
 
440
  ### Supported Tasks and Leaderboards
441
 
442
  - `automatic-speech-recognition`, `speaker-identification`: The dataset can be used to train a model for Automatic Speech Recognition (ASR). The model is presented with an audio file and asked to transcribe the audio file to written text. The most common evaluation metric is the word error rate (WER). The task has an active leaderboard which can be found at https://paperswithcode.com/dataset/multilingual-librispeech and ranks models based on their WER.
443
+ - `text-to-speech`, `text-to-audio`: The dataset can also be used to train a model for Text-To-Speech (TTS).
444
 
445
  ### Languages
446
 
 
453
  For example, to download the German config, simply specify the corresponding language config name (i.e., "german" for German):
454
  ```python
455
  from datasets import load_dataset
 
456
  mls = load_dataset("facebook/multilingual_librispeech", "german", split="train")
457
  ```
458
 
459
  Using the datasets library, you can also stream the dataset on-the-fly by adding a `streaming=True` argument to the `load_dataset` function call. Loading a dataset in streaming mode loads individual samples of the dataset at a time, rather than downloading the entire dataset to disk.
460
  ```python
461
  from datasets import load_dataset
 
462
  mls = load_dataset("facebook/multilingual_librispeech", "german", split="train", streaming=True)
 
463
  print(next(iter(mls)))
464
  ```
465
 
 
470
  ```python
471
  from datasets import load_dataset
472
  from torch.utils.data.sampler import BatchSampler, RandomSampler
 
473
  mls = load_dataset("facebook/multilingual_librispeech", "german", split="train")
474
  batch_sampler = BatchSampler(RandomSampler(mls), batch_size=32, drop_last=False)
475
  dataloader = DataLoader(mls, batch_sampler=batch_sampler)
 
480
  ```python
481
  from datasets import load_dataset
482
  from torch.utils.data import DataLoader
 
483
  mls = load_dataset("facebook/multilingual_librispeech", "german", split="train", streaming=True)
484
  dataloader = DataLoader(mls, batch_size=32)
485
  ```
 
520
  - id: unique id of the data sample.
521
 
522
  - speaker_id: unique id of the speaker. The same speaker id can be found for multiple data samples.
 
523
  - chapter_id: id of the audiobook chapter which includes the transcription.
524
 
525
  ### Data Splits
526
 
527
+ | Number of samples | Train | Train.9h | Train.1h | Dev | Test |
528
  | ----- | ------ | ----- | ---- | ---- | ---- |
529
  | german | 469942 | 2194 | 241 | 3469 | 3394 |
530
  | dutch | 374287 | 2153 | 234 | 3095 | 3075 |
 
534
  | portuguese | 37533 | 2116 | 236 | 826 | 871 |
535
  | polish | 25043 | 2173 | 238 | 512 | 520 |
536
 
 
 
537
  ## Dataset Creation
538
 
539
  ### Curation Rationale
 
600
  }
601
  ```
602
 
603
+
604
+ ### Data Statistics
605
+
606
+ | Duration (h) | Train | Dev | Test |
607
+ |--------------|-----------|-------|-------|
608
+ | English | 44,659.74 | 15.75 | 15.55 |
609
+ | German | 1,966.51 | 14.28 | 14.29 |
610
+ | Dutch | 1,554.24 | 12.76 | 12.76 |
611
+ | French | 1,076.58 | 10.07 | 10.07 |
612
+ | Spanish | 917.68 | 9.99 | 10 |
613
+ | Italian | 247.38 | 5.18 | 5.27 |
614
+ | Portuguese | 160.96 | 3.64 | 3.74 |
615
+ | Polish | 103.65 | 2.08 | 2.14 |
616
+
617
+ | # Speakers | Train | | Dev | | Test | |
618
+ |------------|-------|------|-----|----|------|----|
619
+ | Gender | M | F | M | F | M | F |
620
+ | English | 2742 | 2748 | 21 | 21 | 21 | 21 |
621
+ | German | 81 | 95 | 15 | 15 | 15 | 15 |
622
+ | Dutch | 9 | 31 | 3 | 3 | 3 | 3 |
623
+ | French | 62 | 80 | 9 | 9 | 9 | 9 |
624
+ | Spanish | 36 | 50 | 10 | 10 | 10 | 10 |
625
+ | Italian | 22 | 43 | 5 | 5 | 5 | 5 |
626
+ | Portuguese | 26 | 16 | 5 | 5 | 5 | 5 |
627
+ | Polish | 6 | 5 | 2 | 2 | 2 | 2 |
628
+
629
+ | # Hours / Gender | Dev | | Test | |
630
+ |------------------|------|------|------|------|
631
+ | Gender | M | F | M | F |
632
+ | English | 7.76 | 7.99 | 7.62 | 7.93 |
633
+ | German | 7.06 | 7.22 | 7 | 7.29 |
634
+ | Dutch | 6.44 | 6.32 | 6.72 | 6.04 |
635
+ | French | 5.13 | 4.94 | 5.04 | 5.02 |
636
+ | Spanish | 4.91 | 5.08 | 4.78 | 5.23 |
637
+ | Italian | 2.5 | 2.68 | 2.38 | 2.9 |
638
+ | Portuguese | 1.84 | 1.81 | 1.83 | 1.9 |
639
+ | Polish | 1.12 | 0.95 | 1.09 | 1.05 |
640
+
641
+
642
+
643
+
644
  ### Contributions
645
 
646
+ Thanks to [@patrickvonplaten](https://github.com/patrickvonplaten) and [@polinaeterna](https://github.com/polinaeterna) for adding this dataset.
 
create_dataset.py ADDED
@@ -0,0 +1,106 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ from datasets import DatasetDict, Audio
3
+ import pandas as pd
4
+ from datasets.table import embed_table_storage
5
+ import argparse
6
+
7
+
8
+ if __name__ == "__main__":
9
+ parser = argparse.ArgumentParser()
10
+
11
+
12
+ parser.add_argument("main_folder_path", type=str, help="Path of the base mls folder")
13
+ parser.add_argument("configuration", type=str, help="Dataset configuration to use, if necessary. Here corresponds to the language name.")
14
+ parser.add_argument("output_dir", type=str, help="Save the dataset on disk with this path.")
15
+
16
+ parser.add_argument("--cpu_num_workers", default=1, type=int, help="Number of CPU workers.")
17
+ parser.add_argument("--csv_folder_path", default=None, type=str, help="Path where to save intermediate csv, by default will be main_foldr_path")
18
+ parser.add_argument("--repo_id", default="facebook/multilingual_librispeech", type=str, help="Push the dataset to the hub.")
19
+
20
+
21
+ args = parser.parse_args()
22
+
23
+ main_folder_path = args.main_folder_path
24
+ csv_folder_path = args.csv_folder_path if args.csv_folder_path is not None else main_folder_path
25
+ if not os.path.exists(csv_folder_path):
26
+ os.makedirs(csv_folder_path)
27
+
28
+ splits = ["dev", "test", "train"]
29
+
30
+ # total_length_per_split = 10_000 * 60 * 60 # in sec -> 10k hours
31
+
32
+ csv_dict = {}
33
+ for split in splits:
34
+ segment_path = os.path.join(main_folder_path, split, "segments.txt")
35
+ transcript_path = os.path.join(main_folder_path, split, "transcripts.txt")
36
+
37
+ segments = pd.read_csv(segment_path, sep='\t', names=["audio", "original_path", "begin_time", "end_time"],
38
+ index_col="audio")
39
+ transcripts = pd.read_csv(transcript_path, sep='\t', names=["audio", "transcript"], index_col="audio")
40
+
41
+ df = pd.concat([segments, transcripts], axis=1, join="inner")
42
+ print(
43
+ f"Segments and transcripts of {split} has been joined: new length {len(df)}, old lengths {(len(segments), len(transcripts))}")
44
+
45
+ # add audio duration
46
+ df["audio_duration"] = df["end_time"] - df["begin_time"]
47
+ df["split"] = split
48
+
49
+ print(f"len df {len(df)}")
50
+
51
+ df.to_csv(os.path.join(csv_folder_path, f"{split}.csv"))
52
+ csv_dict[split] = os.path.join(csv_folder_path, f"{split}.csv")
53
+
54
+ # take care of /limited_supervision
55
+ if split == "train":
56
+ nine_hours_segment_path = os.path.join(main_folder_path, "train/limited_supervision/9hr/handles.txt")
57
+ nine_hours_segment = pd.read_csv(nine_hours_segment_path, sep='\t', names=["audio"], index_col="audio").index
58
+ nine_hours_df = df.filter(items=nine_hours_segment, axis=0)
59
+ nine_hours_df.to_csv(os.path.join(csv_folder_path, f"9_hours.csv"))
60
+ csv_dict["9_hours"] = os.path.join(csv_folder_path, f"9_hours.csv")
61
+
62
+ one_hours_segments = [ os.path.join(f.path, "handles.txt") for f in os.scandir( os.path.join(main_folder_path, "train/limited_supervision/1hr")) if f.is_dir()]
63
+ one_hours_segments = pd.concat([pd.read_csv(one, sep='\t', names=["audio"], index_col="audio") for one in one_hours_segments], axis=0).index
64
+ one_hours_df = df.filter(items=one_hours_segments, axis=0)
65
+ one_hours_df.to_csv(os.path.join(csv_folder_path, f"1_hours.csv"))
66
+ csv_dict["1_hours"] = os.path.join(csv_folder_path, f"1_hours.csv")
67
+
68
+
69
+
70
+
71
+ dataset = DatasetDict.from_csv(csv_dict)
72
+
73
+ def extract_speaker_id_and_format_path(audio, split):
74
+ speaker_id = audio.split("_")[0]
75
+ chapter_id = audio.split("_")[1]
76
+ file = f"{audio}.opus"
77
+
78
+ path = os.path.join(main_folder_path, split, "audio", speaker_id, chapter_id, file)
79
+ return {"audio": path, "speaker_id": speaker_id, "chapter_id": chapter_id, "file": file, "id": audio}
80
+
81
+ # correct audio path
82
+ dataset = dataset.map(extract_speaker_id_and_format_path, input_columns=["audio", "split"], num_proc=args.cpu_num_workers, remove_columns=["split"])
83
+ dataset = dataset.cast_column("audio", Audio())
84
+
85
+ print(dataset)
86
+ print(dataset["dev"][0])
87
+
88
+ print("Embed table storage")
89
+
90
+ # load_dataset(...)
91
+ format = dataset["train"].format
92
+ dataset = dataset.with_format("arrow")
93
+ dataset = dataset.map(embed_table_storage, batched=True, num_proc=args.cpu_num_workers)
94
+ dataset = dataset.with_format(**format)
95
+
96
+
97
+ dataset.save_to_disk(args.output_dir, num_proc=args.cpu_num_workers)
98
+
99
+ if args.repo_id:
100
+ pushed = False
101
+ while not pushed:
102
+ try:
103
+ dataset.push_to_hub(args.repo_id, args.configuration, revision="refs/pr/15")
104
+ pushed = True
105
+ except:
106
+ pass
dutch/1_hours-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ba5b95170c0452a71580c730104d42fce130ef291922fa5bec8c94abd4b24bb
3
+ size 15398863
dutch/9_hours-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:173586e5abe761359d5d6906d630f6826a66e41a588945c0add7a8ecdb883c74
3
+ size 139332979
dutch/dev-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0c40d8db30f3cc86041430fb43f091768cba1e65b0b27d8cc3cda02e28571881
3
+ size 198983422
dutch/test-00000-of-00001.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:feb88a282be746e7c14e1f09c017b4f69b07df9c25c94a3943eaac5ca7794ad5
3
+ size 198390898
dutch/train-00000-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d5197c6697455438f0f22a32e8108fd9295b8f21071d46c06f8ef5822e879b1b
3
+ size 498278678
dutch/train-00001-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7e09d94d9eb6f8a8df26ff85377c1a5b1d8db68f5fdd175e6dbcf7610b18e1c6
3
+ size 493730836
dutch/train-00002-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:da9939252200d99e8e1c7922b16b720ee9f0798779ee4446d357de84d4c6c50a
3
+ size 492751421
dutch/train-00003-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:19f5d62017ffd56dfcbd2ddb37d816c81c9c5b0c81b5468f727c9fd659b6e5a1
3
+ size 492439483
dutch/train-00004-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38a9df9aa261cdd5a19ab234552a6de54da5760e681f91f7b43667fc2a169fb2
3
+ size 507283150
dutch/train-00005-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:84e5473fc0d4e1820ace39eb130b843ffc4550fabd53dbdfd3de1adbd52942df
3
+ size 506228652
dutch/train-00006-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bca27f4aad28aeb155544c321205e1c3316791b6492d1e7e4290eff18be54f95
3
+ size 503469896
dutch/train-00007-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3c2977b37bd03ac58acd5227575cafd13604aa42f88fb266a67a7b594fa6e2a
3
+ size 505732565
dutch/train-00008-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c58e2cd6150500777ffd19067db6474419bfeaa8517accc37febd7d79ca0cfb4
3
+ size 497204004
dutch/train-00009-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1045ccd20f04d49acf356c29145c349edf31d5ecb4879b012e62f4b290ccf73f
3
+ size 499720663
dutch/train-00010-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c6131ece5adc923f7a4d78a90c7f537726eb535c684535205ddffb60e5ef0edc
3
+ size 509800583
dutch/train-00011-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a7aee3d7a9d5bf05ca454f85cd1c11244115f1656239cc21ca6d3aa104982b2
3
+ size 496752382
dutch/train-00012-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7bef88920a3c837426536529c50538e92c22a9dde91b3997b061d1a49cb07b4
3
+ size 490141218
dutch/train-00013-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:70f1ccf17d7071dee92a91dcf17ac170d94a7559016070d3ee693538c0251a7c
3
+ size 495052581
dutch/train-00014-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ae9d4a435e0c0316ae50fe2cf5d97317a69b01ed9ca127e6b008433469e4ec13
3
+ size 497074003
dutch/train-00015-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:56a2f5bb81abd080e60a58400a1618dd7deb997089df6592dbbd6757c26742d1
3
+ size 496902151
dutch/train-00016-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3828df35f9fac1e0f3938f652050dd8c1b7ba67438939aae4002fc1986544d8
3
+ size 500414203
dutch/train-00017-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be08e3f22832d9c1855acc6155192bd39a0af5595584bde51e9ce85fd2ef0f1c
3
+ size 497591344
dutch/train-00018-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6e706257981e9558d694407154d1c894d3fb424f2a969c0237f283d06ac6c658
3
+ size 490266080
dutch/train-00019-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2a1e2db14534a9464fca84339b35472ea9d3af9fb7b15084319e71ee7e92087e
3
+ size 490653457
dutch/train-00020-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:61f23455642dadd84b2f968edc5b00ecc674db749dcd12d9a3a38967d52b1500
3
+ size 495251474
dutch/train-00021-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:773ff43f23d2764093b60c9827e1584c0a62bafe14e9c782f38c7083de799590
3
+ size 495863633
dutch/train-00022-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a4ad84fa298641ee71f899e7806b5dac2decfeb714f3be31d1ce01bd450f0255
3
+ size 498791755
dutch/train-00023-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a52e17a2b7e521139c21c623f6be7ca0e176cbad1e8539b4b623da723c33e68f
3
+ size 494115983
dutch/train-00024-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f55d61f3301ef7f59463129a13b2b4bb0e969d616f3250be621a8024a515bc10
3
+ size 496357824
dutch/train-00025-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5bccbdee0d7c3a55faa52caa8d4354f1b53b8f5ae8eb417053b0054930caa8bf
3
+ size 496049208
dutch/train-00026-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d3ef6dc3b3e3c139e98d462ae96fcbf0e0b83011fa1735064fce98ab285a1e2
3
+ size 492055241
dutch/train-00027-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6948db6a68c4a6192f4a8d135aa7c6b9fce89ee09c9bd46ee2437525130905ae
3
+ size 497549729
dutch/train-00028-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:36afd357fae1ec6e042c72e90e39832c92e4f02fe506f1e3c520dd88e5492bb9
3
+ size 491405199
dutch/train-00029-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0b5550bafa417f5c81611125bb5f244b21e40012191b73fc58ffc92b4647d9b5
3
+ size 490596710
dutch/train-00030-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d541892b1ad46a6988c96196d832f5e12b17c81bfb1842121206ad86ed2c9322
3
+ size 492830835
dutch/train-00031-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ceb71076ebc5d3b0b20edd16597d0cdc894e2fb69b9157f5f2156be539275c3b
3
+ size 492267570
dutch/train-00032-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d4b8697914269d883adabdb59746e793efdf166e26b7191b6d83663aeab6b002
3
+ size 498742993
dutch/train-00033-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:27467712ad4a83a8a07dac91f2811db7af6ce1e4fbc65976d5bc701af7941591
3
+ size 489899475
dutch/train-00034-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38ce47c438759c725ca1d71d01e253492d2c3e62975a4df71be313037655c215
3
+ size 495329697
dutch/train-00035-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:00e5af34f05f99d093444b222e2a735a9b339d795b5a10acd5b0754e51bd9645
3
+ size 493591605
dutch/train-00036-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8da20769d1b9422d3cca118b3f0e621dee00ddd86cb824810ecd290e90655b57
3
+ size 495242385
dutch/train-00037-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3163a6f718dabb3c87784fa48d5f7101a078c23e0f0c0a650b2fe175882f420
3
+ size 497938540
dutch/train-00038-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5efa29e0a500ed52f32c84852da8dc74d629d73b91fa99c7aa81dcd67a8044be
3
+ size 493616604
dutch/train-00039-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b9694d9e6eacf8d63e3bad43e4f3cba401caf1215c2cc243923369880c6e7d78
3
+ size 498235792
dutch/train-00040-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:32dcb54c6060b4f69a63cdbf65046d137875d2732af2cf846dfac57f2b813d21
3
+ size 492895497
dutch/train-00041-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f7824d10dc265289c5ed1d1e8b24b7e3c6b1a52aeacf4777521947188956ce0a
3
+ size 495966896
dutch/train-00042-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87f323c63bcbfd02ab606dfafaef0d01f9a16591dfbf45ed8ee5ea2702c9e1a3
3
+ size 495535322
dutch/train-00043-of-00048.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7c9b9319fa8870c4befedd1d4535bf4e488719b598662ba75bbed78fbd0a81f0
3
+ size 495532373