jonatasgrosman commited on
Commit
54bcf04
1 Parent(s): 87bcbd1

jonatasgrosman/exp_w2v2r_de_xls-r_age_teens-8_sixties-2_s136

Browse files
.gitattributes CHANGED
@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
32
  *.zip filter=lfs diff=lfs merge=lfs -text
33
  *.zst filter=lfs diff=lfs merge=lfs -text
34
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
32
  *.zip filter=lfs diff=lfs merge=lfs -text
33
  *.zst filter=lfs diff=lfs merge=lfs -text
34
  *tfevents* filter=lfs diff=lfs merge=lfs -text
35
+ cv7_transcriptions.json filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,17 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - de
4
+ license: apache-2.0
5
+ tags:
6
+ - automatic-speech-recognition
7
+ - de
8
+ datasets:
9
+ - mozilla-foundation/common_voice_7_0
10
+ ---
11
+ # exp_w2v2r_de_xls-r_age_teens-8_sixties-2_s136
12
+
13
+ Fine-tuned [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) for speech recognition using the train split of [Common Voice 7.0 (de)](https://huggingface.co/datasets/mozilla-foundation/common_voice_7_0).
14
+ When using this model, make sure that your speech input is sampled at 16kHz.
15
+
16
+ This model has been fine-tuned by the [HuggingSound](https://github.com/jonatasgrosman/huggingsound) tool.
17
+
config.json ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "facebook/wav2vec2-xls-r-300m",
3
+ "activation_dropout": 0.05,
4
+ "adapter_kernel_size": 3,
5
+ "adapter_stride": 2,
6
+ "add_adapter": false,
7
+ "apply_spec_augment": true,
8
+ "architectures": [
9
+ "Wav2Vec2ForCTC"
10
+ ],
11
+ "attention_dropout": 0.05,
12
+ "bos_token_id": 1,
13
+ "classifier_proj_size": 256,
14
+ "codevector_dim": 768,
15
+ "contrastive_logits_temperature": 0.1,
16
+ "conv_bias": true,
17
+ "conv_dim": [
18
+ 512,
19
+ 512,
20
+ 512,
21
+ 512,
22
+ 512,
23
+ 512,
24
+ 512
25
+ ],
26
+ "conv_kernel": [
27
+ 10,
28
+ 3,
29
+ 3,
30
+ 3,
31
+ 3,
32
+ 2,
33
+ 2
34
+ ],
35
+ "conv_stride": [
36
+ 5,
37
+ 2,
38
+ 2,
39
+ 2,
40
+ 2,
41
+ 2,
42
+ 2
43
+ ],
44
+ "ctc_loss_reduction": "mean",
45
+ "ctc_zero_infinity": false,
46
+ "diversity_loss_weight": 0.1,
47
+ "do_stable_layer_norm": true,
48
+ "eos_token_id": 2,
49
+ "feat_extract_activation": "gelu",
50
+ "feat_extract_dropout": 0.0,
51
+ "feat_extract_norm": "layer",
52
+ "feat_proj_dropout": 0.05,
53
+ "feat_quantizer_dropout": 0.0,
54
+ "final_dropout": 0.05,
55
+ "gradient_checkpointing": false,
56
+ "hidden_act": "gelu",
57
+ "hidden_dropout": 0.05,
58
+ "hidden_size": 1024,
59
+ "initializer_range": 0.02,
60
+ "intermediate_size": 4096,
61
+ "layer_norm_eps": 1e-05,
62
+ "layerdrop": 0.05,
63
+ "mask_feature_length": 10,
64
+ "mask_feature_min_masks": 0,
65
+ "mask_feature_prob": 0.0,
66
+ "mask_time_length": 10,
67
+ "mask_time_min_masks": 2,
68
+ "mask_time_prob": 0.05,
69
+ "model_type": "wav2vec2",
70
+ "num_adapter_layers": 3,
71
+ "num_attention_heads": 16,
72
+ "num_codevector_groups": 2,
73
+ "num_codevectors_per_group": 320,
74
+ "num_conv_pos_embedding_groups": 16,
75
+ "num_conv_pos_embeddings": 128,
76
+ "num_feat_extract_layers": 7,
77
+ "num_hidden_layers": 24,
78
+ "num_negatives": 100,
79
+ "output_hidden_size": 1024,
80
+ "pad_token_id": 0,
81
+ "proj_codevector_dim": 768,
82
+ "tdnn_dilation": [
83
+ 1,
84
+ 2,
85
+ 3,
86
+ 1,
87
+ 1
88
+ ],
89
+ "tdnn_dim": [
90
+ 512,
91
+ 512,
92
+ 512,
93
+ 512,
94
+ 1500
95
+ ],
96
+ "tdnn_kernel": [
97
+ 5,
98
+ 3,
99
+ 3,
100
+ 1,
101
+ 1
102
+ ],
103
+ "torch_dtype": "float32",
104
+ "transformers_version": "4.18.0",
105
+ "use_weighted_layer_sum": false,
106
+ "vocab_size": 38,
107
+ "xvector_output_dim": 512
108
+ }
cv7_evaluation.json ADDED
@@ -0,0 +1,396 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "validation": [
3
+ {
4
+ "subset_type": "all",
5
+ "subset": "all",
6
+ "cer": 0.04492097390914287,
7
+ "wer": 0.18271583124932042
8
+ },
9
+ {
10
+ "subset_type": "age",
11
+ "subset": "",
12
+ "cer": 0.04427924472032959,
13
+ "wer": 0.18112412684933105
14
+ },
15
+ {
16
+ "subset_type": "age",
17
+ "subset": "twenties",
18
+ "cer": 0.047698760163349396,
19
+ "wer": 0.1942563330825182
20
+ },
21
+ {
22
+ "subset_type": "age",
23
+ "subset": "thirties",
24
+ "cer": 0.04144505770195685,
25
+ "wer": 0.16447368421052633
26
+ },
27
+ {
28
+ "subset_type": "age",
29
+ "subset": "teens",
30
+ "cer": 0.04390695647436945,
31
+ "wer": 0.18682545589877186
32
+ },
33
+ {
34
+ "subset_type": "age",
35
+ "subset": "seventies",
36
+ "cer": 0.0432113920942794,
37
+ "wer": 0.195578231292517
38
+ },
39
+ {
40
+ "subset_type": "age",
41
+ "subset": "fourties",
42
+ "cer": 0.06328320802005012,
43
+ "wer": 0.22857142857142856
44
+ },
45
+ {
46
+ "subset_type": "age",
47
+ "subset": "eighties",
48
+ "cer": 0.05057096247960848,
49
+ "wer": 0.23170731707317074
50
+ },
51
+ {
52
+ "subset_type": "age",
53
+ "subset": "fifties",
54
+ "cer": 0.04711268645694875,
55
+ "wer": 0.18745785569790965
56
+ },
57
+ {
58
+ "subset_type": "age",
59
+ "subset": "sixties",
60
+ "cer": 0.051874907366236844,
61
+ "wer": 0.20020909566126502
62
+ },
63
+ {
64
+ "subset_type": "gender",
65
+ "subset": "",
66
+ "cer": 0.04427881875049604,
67
+ "wer": 0.18112563483473482
68
+ },
69
+ {
70
+ "subset_type": "gender",
71
+ "subset": "female",
72
+ "cer": 0.04561535566428303,
73
+ "wer": 0.1796267496111975
74
+ },
75
+ {
76
+ "subset_type": "gender",
77
+ "subset": "other",
78
+ "cer": 0.0765639589169001,
79
+ "wer": 0.2804878048780488
80
+ },
81
+ {
82
+ "subset_type": "gender",
83
+ "subset": "male",
84
+ "cer": 0.04798255873080443,
85
+ "wer": 0.19083405734144224
86
+ },
87
+ {
88
+ "subset_type": "accent",
89
+ "subset": "",
90
+ "cer": 0.04483434875755759,
91
+ "wer": 0.18282481699036537
92
+ },
93
+ {
94
+ "subset_type": "accent",
95
+ "subset": "canada",
96
+ "cer": 0.08455882352941177,
97
+ "wer": 0.325
98
+ },
99
+ {
100
+ "subset_type": "accent",
101
+ "subset": "france",
102
+ "cer": 0.05067567567567568,
103
+ "wer": 0.22916666666666666
104
+ },
105
+ {
106
+ "subset_type": "accent",
107
+ "subset": "belgium",
108
+ "cer": 0.05405405405405406,
109
+ "wer": 0.2692307692307692
110
+ },
111
+ {
112
+ "subset_type": "accent",
113
+ "subset": "slovakia",
114
+ "cer": 0.04722222222222222,
115
+ "wer": 0.22807017543859648
116
+ },
117
+ {
118
+ "subset_type": "accent",
119
+ "subset": "netherlands",
120
+ "cer": 0.05723905723905724,
121
+ "wer": 0.21739130434782608
122
+ },
123
+ {
124
+ "subset_type": "accent",
125
+ "subset": "switzerland",
126
+ "cer": 0.05278263039680559,
127
+ "wer": 0.2141623488773748
128
+ },
129
+ {
130
+ "subset_type": "accent",
131
+ "subset": "luxembourg",
132
+ "cer": 0.08353510895883777,
133
+ "wer": 0.3140495867768595
134
+ },
135
+ {
136
+ "subset_type": "accent",
137
+ "subset": "czechia",
138
+ "cer": 0.038527397260273974,
139
+ "wer": 0.14545454545454545
140
+ },
141
+ {
142
+ "subset_type": "accent",
143
+ "subset": "turkey",
144
+ "cer": 0.13022113022113022,
145
+ "wer": 0.4915254237288136
146
+ },
147
+ {
148
+ "subset_type": "accent",
149
+ "subset": "russia",
150
+ "cer": 0.0674061433447099,
151
+ "wer": 0.24064171122994651
152
+ },
153
+ {
154
+ "subset_type": "accent",
155
+ "subset": "austria",
156
+ "cer": 0.04820165094339623,
157
+ "wer": 0.19238578680203045
158
+ },
159
+ {
160
+ "subset_type": "accent",
161
+ "subset": "poland",
162
+ "cer": 0.04390243902439024,
163
+ "wer": 0.18181818181818182
164
+ },
165
+ {
166
+ "subset_type": "accent",
167
+ "subset": "united_kingdom",
168
+ "cer": 0.08549222797927461,
169
+ "wer": 0.3162393162393162
170
+ },
171
+ {
172
+ "subset_type": "accent",
173
+ "subset": "other",
174
+ "cer": 0.05748502994011976,
175
+ "wer": 0.216
176
+ },
177
+ {
178
+ "subset_type": "accent",
179
+ "subset": "germany",
180
+ "cer": 0.04285577551964522,
181
+ "wer": 0.1729335672001754
182
+ },
183
+ {
184
+ "subset_type": "accent",
185
+ "subset": "united_states",
186
+ "cer": 0.0861797162375197,
187
+ "wer": 0.303886925795053
188
+ }
189
+ ],
190
+ "test": [
191
+ {
192
+ "subset_type": "all",
193
+ "subset": "all",
194
+ "cer": 0.05115450300694033,
195
+ "wer": 0.1985494106980961
196
+ },
197
+ {
198
+ "subset_type": "age",
199
+ "subset": "",
200
+ "cer": 0.05031743978013447,
201
+ "wer": 0.1961445598436362
202
+ },
203
+ {
204
+ "subset_type": "age",
205
+ "subset": "twenties",
206
+ "cer": 0.053694690969240066,
207
+ "wer": 0.21451717433570966
208
+ },
209
+ {
210
+ "subset_type": "age",
211
+ "subset": "fifties",
212
+ "cer": 0.046683399067766224,
213
+ "wer": 0.19602977667493796
214
+ },
215
+ {
216
+ "subset_type": "age",
217
+ "subset": "teens",
218
+ "cer": 0.0642579129029889,
219
+ "wer": 0.23496993987975953
220
+ },
221
+ {
222
+ "subset_type": "age",
223
+ "subset": "seventies",
224
+ "cer": 0.06611368626494267,
225
+ "wer": 0.2571912013536379
226
+ },
227
+ {
228
+ "subset_type": "age",
229
+ "subset": "fourties",
230
+ "cer": 0.07876517064429421,
231
+ "wer": 0.25041551246537397
232
+ },
233
+ {
234
+ "subset_type": "age",
235
+ "subset": "eighties",
236
+ "cer": 0.06788990825688074,
237
+ "wer": 0.2876712328767123
238
+ },
239
+ {
240
+ "subset_type": "age",
241
+ "subset": "thirties",
242
+ "cer": 0.05276852147715342,
243
+ "wer": 0.19738442822384428
244
+ },
245
+ {
246
+ "subset_type": "age",
247
+ "subset": "sixties",
248
+ "cer": 0.07248348231636223,
249
+ "wer": 0.2547683923705722
250
+ },
251
+ {
252
+ "subset_type": "gender",
253
+ "subset": "",
254
+ "cer": 0.05039616258610655,
255
+ "wer": 0.19636973811238642
256
+ },
257
+ {
258
+ "subset_type": "gender",
259
+ "subset": "female",
260
+ "cer": 0.06805109886783314,
261
+ "wer": 0.2554438860971524
262
+ },
263
+ {
264
+ "subset_type": "gender",
265
+ "subset": "other",
266
+ "cer": 0.1245819397993311,
267
+ "wer": 0.2556818181818182
268
+ },
269
+ {
270
+ "subset_type": "gender",
271
+ "subset": "male",
272
+ "cer": 0.054841443604774345,
273
+ "wer": 0.20961814986419555
274
+ },
275
+ {
276
+ "subset_type": "accent",
277
+ "subset": "",
278
+ "cer": 0.051013207868536255,
279
+ "wer": 0.19819005212528304
280
+ },
281
+ {
282
+ "subset_type": "accent",
283
+ "subset": "austria",
284
+ "cer": 0.05869682283252558,
285
+ "wer": 0.22437673130193905
286
+ },
287
+ {
288
+ "subset_type": "accent",
289
+ "subset": "united_states",
290
+ "cer": 0.0743651753325272,
291
+ "wer": 0.24066390041493776
292
+ },
293
+ {
294
+ "subset_type": "accent",
295
+ "subset": "belgium",
296
+ "cer": 0.07614213197969544,
297
+ "wer": 0.38095238095238093
298
+ },
299
+ {
300
+ "subset_type": "accent",
301
+ "subset": "luxembourg",
302
+ "cer": 0.045774647887323945,
303
+ "wer": 0.18421052631578946
304
+ },
305
+ {
306
+ "subset_type": "accent",
307
+ "subset": "denmark",
308
+ "cer": 0.07352941176470588,
309
+ "wer": 0.4444444444444444
310
+ },
311
+ {
312
+ "subset_type": "accent",
313
+ "subset": "czechia",
314
+ "cer": 0.15176151761517614,
315
+ "wer": 0.4716981132075472
316
+ },
317
+ {
318
+ "subset_type": "accent",
319
+ "subset": "united_kingdom",
320
+ "cer": 0.11442006269592477,
321
+ "wer": 0.35294117647058826
322
+ },
323
+ {
324
+ "subset_type": "accent",
325
+ "subset": "france",
326
+ "cer": 0.07856450048496605,
327
+ "wer": 0.28
328
+ },
329
+ {
330
+ "subset_type": "accent",
331
+ "subset": "slovakia",
332
+ "cer": 0.047979797979797977,
333
+ "wer": 0.1896551724137931
334
+ },
335
+ {
336
+ "subset_type": "accent",
337
+ "subset": "bulgaria",
338
+ "cer": 0.0,
339
+ "wer": 0.0
340
+ },
341
+ {
342
+ "subset_type": "accent",
343
+ "subset": "russia",
344
+ "cer": 0.16835016835016836,
345
+ "wer": 0.4666666666666667
346
+ },
347
+ {
348
+ "subset_type": "accent",
349
+ "subset": "greece",
350
+ "cer": 0.09615384615384616,
351
+ "wer": 0.2222222222222222
352
+ },
353
+ {
354
+ "subset_type": "accent",
355
+ "subset": "netherlands",
356
+ "cer": 0.06429652042360061,
357
+ "wer": 0.26424870466321243
358
+ },
359
+ {
360
+ "subset_type": "accent",
361
+ "subset": "switzerland",
362
+ "cer": 0.06454231784047088,
363
+ "wer": 0.2338935574229692
364
+ },
365
+ {
366
+ "subset_type": "accent",
367
+ "subset": "turkey",
368
+ "cer": 0.1935483870967742,
369
+ "wer": 0.4
370
+ },
371
+ {
372
+ "subset_type": "accent",
373
+ "subset": "poland",
374
+ "cer": 0.09866220735785954,
375
+ "wer": 0.32608695652173914
376
+ },
377
+ {
378
+ "subset_type": "accent",
379
+ "subset": "other",
380
+ "cer": 0.0,
381
+ "wer": 0.0
382
+ },
383
+ {
384
+ "subset_type": "accent",
385
+ "subset": "germany",
386
+ "cer": 0.04803580176625578,
387
+ "wer": 0.18911123129619614
388
+ },
389
+ {
390
+ "subset_type": "accent",
391
+ "subset": "hungary",
392
+ "cer": 0.06153846153846154,
393
+ "wer": 0.42857142857142855
394
+ }
395
+ ]
396
+ }
cv7_transcriptions.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:03f719b8bfa255a880e980c91acbc8c8d49248ee3ab3f5f3a20c04f78f9f8f82
3
+ size 138771755
preprocessor_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0,
7
+ "processor_class": "Wav2Vec2Processor",
8
+ "return_attention_mask": true,
9
+ "sampling_rate": 16000
10
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb8476e7705919db4544f6d9aa9caf7c515e0a8ec17e1fe4a7f7f2293dd084e8
3
+ size 1262054513
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "<pad>", "do_lower_case": false, "word_delimiter_token": "|", "replace_word_delimiter_char": " ", "special_tokens_map_file": null, "name_or_path": "/tmp/tmpxe2f7vu4", "tokenizer_class": "Wav2Vec2CTCTokenizer", "processor_class": "Wav2Vec2Processor"}
vocab.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"<pad>": 0, "<s>": 1, "</s>": 2, "<unk>": 3, "|": 4, "'": 5, "-": 6, "a": 7, "b": 8, "c": 9, "d": 10, "e": 11, "f": 12, "g": 13, "h": 14, "i": 15, "j": 16, "k": 17, "l": 18, "m": 19, "n": 20, "o": 21, "p": 22, "q": 23, "r": 24, "s": 25, "t": 26, "u": 27, "v": 28, "w": 29, "x": 30, "y": 31, "z": 32, "ß": 33, "á": 34, "ä": 35, "ö": 36, "ü": 37}