Hugging Face's logo

hf-audio
/
xcodec-wavlm-mls

Feature Extraction
Transformers
Safetensors
xcodec
Model card Files
xet
Community
bezzam HF Staff commited on
Commit
d8876e8
·
verified ·
1 Parent(s): 668299e

Upload model

Browse files
Files changed (2) hide show
  1. config.json +181 -0
  2. model.safetensors +3 -0
config.json ADDED
@@ -0,0 +1,181 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "acoustic_model_config": {
3
+ "codebook_dim": 8,
4
+ "codebook_loss_weight": 1.0,
5
+ "codebook_size": 1024,
6
+ "commitment_loss_weight": 0.25,
7
+ "decoder_hidden_size": 1024,
8
+ "downsampling_ratios": [
9
+ 8,
10
+ 5,
11
+ 4,
12
+ 2
13
+ ],
14
+ "encoder_hidden_size": 64,
15
+ "hidden_size": 256,
16
+ "hop_length": 320,
17
+ "model_type": "dac",
18
+ "n_codebooks": 9,
19
+ "quantizer_dropout": 0,
20
+ "sampling_rate": 16000,
21
+ "upsampling_ratios": [
22
+ 8,
23
+ 5,
24
+ 4,
25
+ 2
26
+ ]
27
+ },
28
+ "architectures": [
29
+ "XcodecModel"
30
+ ],
31
+ "block_dilations": [
32
+ 1,
33
+ 1
34
+ ],
35
+ "channel_ratios": [
36
+ 1,
37
+ 1
38
+ ],
39
+ "codebook_dim": 1024,
40
+ "codebook_size": 1024,
41
+ "decoder_channels": 768,
42
+ "encoder_channels": 768,
43
+ "initializer_range": 0.02,
44
+ "input_channels": 768,
45
+ "kernel_size": 3,
46
+ "model_type": "xcodec",
47
+ "output_channels": 768,
48
+ "sample_rate": 16000,
49
+ "semantic_model_config": {
50
+ "_name_or_path": "microsoft/wavlm-base-plus",
51
+ "activation_dropout": 0.0,
52
+ "adapter_kernel_size": 3,
53
+ "adapter_stride": 2,
54
+ "add_adapter": false,
55
+ "apply_spec_augment": true,
56
+ "architectures": [
57
+ "WavLMModel"
58
+ ],
59
+ "attention_dropout": 0.1,
60
+ "classifier_proj_size": 256,
61
+ "codevector_dim": 256,
62
+ "contrastive_logits_temperature": 0.1,
63
+ "conv_bias": false,
64
+ "conv_dim": [
65
+ 512,
66
+ 512,
67
+ 512,
68
+ 512,
69
+ 512,
70
+ 512,
71
+ 512
72
+ ],
73
+ "conv_kernel": [
74
+ 10,
75
+ 3,
76
+ 3,
77
+ 3,
78
+ 3,
79
+ 2,
80
+ 2
81
+ ],
82
+ "conv_stride": [
83
+ 5,
84
+ 2,
85
+ 2,
86
+ 2,
87
+ 2,
88
+ 2,
89
+ 2
90
+ ],
91
+ "ctc_loss_reduction": "sum",
92
+ "ctc_zero_infinity": false,
93
+ "diversity_loss_weight": 0.1,
94
+ "do_stable_layer_norm": false,
95
+ "feat_extract_activation": "gelu",
96
+ "feat_extract_norm": "group",
97
+ "feat_proj_dropout": 0.1,
98
+ "feat_quantizer_dropout": 0.0,
99
+ "final_dropout": 0.0,
100
+ "freeze_feat_extract_train": true,
101
+ "hidden_act": "gelu",
102
+ "hidden_dropout": 0.1,
103
+ "hidden_size": 768,
104
+ "initializer_range": 0.02,
105
+ "intermediate_size": 3072,
106
+ "layer_norm_eps": 1e-05,
107
+ "layerdrop": 0.05,
108
+ "mask_channel_length": 10,
109
+ "mask_channel_min_space": 1,
110
+ "mask_channel_other": 0.0,
111
+ "mask_channel_prob": 0.0,
112
+ "mask_channel_selection": "static",
113
+ "mask_feature_length": 10,
114
+ "mask_feature_min_masks": 0,
115
+ "mask_feature_prob": 0.0,
116
+ "mask_time_length": 10,
117
+ "mask_time_min_masks": 2,
118
+ "mask_time_min_space": 1,
119
+ "mask_time_other": 0.0,
120
+ "mask_time_prob": 0.05,
121
+ "mask_time_selection": "static",
122
+ "max_bucket_distance": 800,
123
+ "model_type": "wavlm",
124
+ "no_mask_channel_overlap": false,
125
+ "no_mask_time_overlap": false,
126
+ "num_adapter_layers": 3,
127
+ "num_attention_heads": 12,
128
+ "num_buckets": 320,
129
+ "num_codevector_groups": 2,
130
+ "num_codevectors_per_group": 320,
131
+ "num_conv_pos_embedding_groups": 16,
132
+ "num_conv_pos_embeddings": 128,
133
+ "num_ctc_classes": 80,
134
+ "num_feat_extract_layers": 7,
135
+ "num_hidden_layers": 12,
136
+ "num_negatives": 100,
137
+ "output_hidden_size": 768,
138
+ "proj_codevector_dim": 256,
139
+ "replace_prob": 0.5,
140
+ "tdnn_dilation": [
141
+ 1,
142
+ 2,
143
+ 3,
144
+ 1,
145
+ 1
146
+ ],
147
+ "tdnn_dim": [
148
+ 512,
149
+ 512,
150
+ 512,
151
+ 512,
152
+ 1500
153
+ ],
154
+ "tdnn_kernel": [
155
+ 5,
156
+ 3,
157
+ 3,
158
+ 1,
159
+ 1
160
+ ],
161
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
162
+ "torch_dtype": "float32",
163
+ "use_weighted_layer_sum": false,
164
+ "vocab_size": 32,
165
+ "xvector_output_dim": 512
166
+ },
167
+ "strides": [
168
+ 1,
169
+ 1
170
+ ],
171
+ "target_bandwidths": [
172
+ 0.5,
173
+ 1,
174
+ 1.5,
175
+ 2,
176
+ 4
177
+ ],
178
+ "torch_dtype": "float32",
179
+ "transformers_version": "4.56.0.dev0",
180
+ "unit_kernel_size": 3
181
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9ae25e265e56ad9916a4ff972a6262b8d03081cb10524795e0b83d9f8219af59
3
+ size 710037116