patrickvonplaten commited on
Commit
99d84ad
1 Parent(s): 48dd949
Files changed (3) hide show
  1. README.md +26 -0
  2. config.json +115 -0
  3. pytorch_model.bin +3 -0
README.md ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language: baltic
3
+ tags:
4
+ - audio
5
+ - automatic-speech-recognition
6
+ - voxpopuli-v2
7
+ datasets:
8
+ - voxpopuli
9
+ license: cc-by-nc-4.0
10
+ inference: false
11
+ ---
12
+
13
+ # Wav2Vec2-large-VoxPopuli-V2
14
+
15
+ [Facebook's Wav2Vec2](https://ai.facebook.com/blog/wav2vec-20-learning-the-structure-of-speech-from-raw-audio/) large model pretrained only in **baltic** on **27.5** unlabeled datat of the [VoxPopuli corpus](https://arxiv.org/abs/2101.00390).
16
+
17
+ The model is pretrained on 16kHz sampled speech audio. When using the model make sure that your speech input is also sampled at 16Khz.
18
+
19
+ **Note**: This model does not have a tokenizer as it was pretrained on audio alone. In order to use this model for **speech recognition**, a tokenizer should be created and the model should be fine-tuned on labeled text data in **baltic**. Check out [this blog](https://huggingface.co/blog/fine-tune-xlsr-wav2vec2) for a more in-detail explanation of how to fine-tune the model.
20
+
21
+ **Paper**: *[VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation
22
+ Learning, Semi-Supervised Learning and Interpretation](https://arxiv.org/abs/2101.00390)*
23
+
24
+ **Authors**: *Changhan Wang, Morgane Riviere, Ann Lee, Anne Wu, Chaitanya Talnikar, Daniel Haziza, Mary Williamson, Juan Pino, Emmanuel Dupoux* from *Facebook AI*.
25
+
26
+ See the official website for more information, [here](https://github.com/facebookresearch/voxpopuli/).
config.json ADDED
@@ -0,0 +1,115 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_dropout": 0.0,
3
+ "adapter_kernel_size": 3,
4
+ "adapter_stride": 2,
5
+ "add_adapter": false,
6
+ "apply_spec_augment": true,
7
+ "architectures": [
8
+ "Wav2Vec2ForPreTraining"
9
+ ],
10
+ "attention_dropout": 0.1,
11
+ "bos_token_id": 1,
12
+ "classifier_proj_size": 256,
13
+ "codevector_dim": 768,
14
+ "contrastive_logits_temperature": 0.1,
15
+ "conv_bias": true,
16
+ "conv_dim": [
17
+ 512,
18
+ 512,
19
+ 512,
20
+ 512,
21
+ 512,
22
+ 512,
23
+ 512
24
+ ],
25
+ "conv_kernel": [
26
+ 10,
27
+ 3,
28
+ 3,
29
+ 3,
30
+ 3,
31
+ 2,
32
+ 2
33
+ ],
34
+ "conv_stride": [
35
+ 5,
36
+ 2,
37
+ 2,
38
+ 2,
39
+ 2,
40
+ 2,
41
+ 2
42
+ ],
43
+ "ctc_loss_reduction": "sum",
44
+ "ctc_zero_infinity": false,
45
+ "diversity_loss_weight": 0.1,
46
+ "do_stable_layer_norm": true,
47
+ "eos_token_id": 2,
48
+ "feat_extract_activation": "gelu",
49
+ "feat_extract_dropout": 0.0,
50
+ "feat_extract_norm": "layer",
51
+ "feat_proj_dropout": 0.1,
52
+ "feat_quantizer_dropout": 0.0,
53
+ "final_dropout": 0.0,
54
+ "gradient_checkpointing": false,
55
+ "hidden_act": "gelu",
56
+ "hidden_dropout": 0.1,
57
+ "hidden_size": 1024,
58
+ "initializer_range": 0.02,
59
+ "intermediate_size": 4096,
60
+ "layer_norm_eps": 1e-05,
61
+ "layerdrop": 0.1,
62
+ "mask_channel_length": 10,
63
+ "mask_channel_min_space": 1,
64
+ "mask_channel_other": 0.0,
65
+ "mask_channel_prob": 0.0,
66
+ "mask_channel_selection": "static",
67
+ "mask_feature_length": 10,
68
+ "mask_feature_min_masks": 0,
69
+ "mask_feature_prob": 0.0,
70
+ "mask_time_length": 10,
71
+ "mask_time_min_masks": 2,
72
+ "mask_time_min_space": 1,
73
+ "mask_time_other": 0.0,
74
+ "mask_time_prob": 0.075,
75
+ "mask_time_selection": "static",
76
+ "model_type": "wav2vec2",
77
+ "num_adapter_layers": 3,
78
+ "num_attention_heads": 16,
79
+ "num_codevector_groups": 2,
80
+ "num_codevectors_per_group": 320,
81
+ "num_conv_pos_embedding_groups": 16,
82
+ "num_conv_pos_embeddings": 128,
83
+ "num_feat_extract_layers": 7,
84
+ "num_hidden_layers": 24,
85
+ "num_negatives": 100,
86
+ "output_hidden_size": 1024,
87
+ "pad_token_id": 0,
88
+ "proj_codevector_dim": 768,
89
+ "tdnn_dilation": [
90
+ 1,
91
+ 2,
92
+ 3,
93
+ 1,
94
+ 1
95
+ ],
96
+ "tdnn_dim": [
97
+ 512,
98
+ 512,
99
+ 512,
100
+ 512,
101
+ 1500
102
+ ],
103
+ "tdnn_kernel": [
104
+ 5,
105
+ 3,
106
+ 3,
107
+ 1,
108
+ 1
109
+ ],
110
+ "torch_dtype": "float32",
111
+ "transformers_version": "4.17.0.dev0",
112
+ "use_weighted_layer_sum": false,
113
+ "vocab_size": 32,
114
+ "xvector_output_dim": 512
115
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bd79eb545c0bc1d36acf109d1d4953c67e7fa77656aec04e2aef34a78c7a0f75
3
+ size 1269729511