Hugging Face's logo

Vansh180
/
deepfake-audio-wav2vec2

Audio Classification
Transformers
Safetensors
PyTorch
English
wav2vec2
deepfake-detection
spoof-detection
Eval Results (legacy)
Model card Files
xet
 Community
Vansh180 commited on
Commit
195cc2d
·
verified ·
1 Parent(s): fe97be1

Upload folder using huggingface_hub

Browse files
Files changed (5) hide show
  1. README.md +61 -0
  2. config.json +125 -0
  3. label_map.json +10 -0
  4. model.safetensors +3 -0
  5. preprocessor_config.json +9 -0
README.md ADDED
@@ -0,0 +1,61 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language: en
3
+ license: mit
4
+ pipeline_tag: audio-classification
5
+ tags:
6
+ - audio-classification
7
+ - deepfake-detection
8
+ - spoof-detection
9
+ - asvspoof
10
+ - wav2vec2
11
+ - pytorch
12
+ - transformers
13
+ metrics:
14
+ - accuracy
15
+ - precision
16
+ - recall
17
+ - f1
18
+ model-index:
19
+ - name: Deepfake-Audio-Wav2Vec2
20
+ results:
21
+ - task:
22
+ type: audio-classification
23
+ name: Audio Classification
24
+ dataset:
25
+ name: Balanced ASVspoof 2021 PA
26
+ type: custom
27
+ metrics:
28
+ - type: accuracy
29
+ value: 0.768
30
+ name: Accuracy
31
+ - type: precision
32
+ value: 0.767
33
+ name: Precision
34
+ - type: recall
35
+ value: 0.768
36
+ name: Recall
37
+ - type: f1
38
+ value: 0.764
39
+ name: F1 Score
40
+ ---
41
+
42
+ # Model Card — Deepfake-Audio-Wav2Vec2
43
+
44
+ ## Model Overview
45
+
46
+ Deepfake-Audio-Wav2Vec2 is a fine-tuned audio classification model trained to detect real vs spoofed audio.
47
+ It is built on top of `facebook/wav2vec2-base` and adapted for binary deepfake audio detection.
48
+
49
+ ## Labels
50
+
51
+ - `0` → bonafide
52
+ - `1` → spoof
53
+
54
+ ## Evaluation Metrics
55
+
56
+ | Metric | Score |
57
+ |------|------:|
58
+ | Accuracy | 76.8% |
59
+ | Precision | 76.7% |
60
+ | Recall | 76.8% |
61
+ | F1 Score | 76.4% |
config.json ADDED
@@ -0,0 +1,125 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation_dropout": 0.0,
3
+ "adapter_attn_dim": null,
4
+ "adapter_kernel_size": 3,
5
+ "adapter_stride": 2,
6
+ "add_adapter": false,
7
+ "apply_spec_augment": true,
8
+ "architectures": [
9
+ "Wav2Vec2ForSequenceClassification"
10
+ ],
11
+ "attention_dropout": 0.1,
12
+ "bos_token_id": 1,
13
+ "classifier_proj_size": 256,
14
+ "codevector_dim": 256,
15
+ "contrastive_logits_temperature": 0.1,
16
+ "conv_bias": false,
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": "sum",
45
+ "ctc_zero_infinity": false,
46
+ "diversity_loss_weight": 0.1,
47
+ "do_stable_layer_norm": false,
48
+ "dtype": "float32",
49
+ "eos_token_id": 2,
50
+ "feat_extract_activation": "gelu",
51
+ "feat_extract_norm": "group",
52
+ "feat_proj_dropout": 0.1,
53
+ "feat_quantizer_dropout": 0.0,
54
+ "final_dropout": 0.0,
55
+ "freeze_feat_extract_train": true,
56
+ "hidden_act": "gelu",
57
+ "hidden_dropout": 0.1,
58
+ "hidden_size": 768,
59
+ "id2label": {
60
+ "0": "real",
61
+ "1": "fake"
62
+ },
63
+ "initializer_range": 0.02,
64
+ "intermediate_size": 3072,
65
+ "label2id": {
66
+ "fake": 1,
67
+ "real": 0
68
+ },
69
+ "layer_norm_eps": 1e-05,
70
+ "layerdrop": 0.0,
71
+ "mask_channel_length": 10,
72
+ "mask_channel_min_space": 1,
73
+ "mask_channel_other": 0.0,
74
+ "mask_channel_prob": 0.0,
75
+ "mask_channel_selection": "static",
76
+ "mask_feature_length": 10,
77
+ "mask_feature_min_masks": 0,
78
+ "mask_feature_prob": 0.0,
79
+ "mask_time_length": 10,
80
+ "mask_time_min_masks": 2,
81
+ "mask_time_min_space": 1,
82
+ "mask_time_other": 0.0,
83
+ "mask_time_prob": 0.05,
84
+ "mask_time_selection": "static",
85
+ "model_type": "wav2vec2",
86
+ "no_mask_channel_overlap": false,
87
+ "no_mask_time_overlap": false,
88
+ "num_adapter_layers": 3,
89
+ "num_attention_heads": 12,
90
+ "num_codevector_groups": 2,
91
+ "num_codevectors_per_group": 320,
92
+ "num_conv_pos_embedding_groups": 16,
93
+ "num_conv_pos_embeddings": 128,
94
+ "num_feat_extract_layers": 7,
95
+ "num_hidden_layers": 12,
96
+ "num_negatives": 100,
97
+ "output_hidden_size": 768,
98
+ "pad_token_id": 0,
99
+ "proj_codevector_dim": 256,
100
+ "tdnn_dilation": [
101
+ 1,
102
+ 2,
103
+ 3,
104
+ 1,
105
+ 1
106
+ ],
107
+ "tdnn_dim": [
108
+ 512,
109
+ 512,
110
+ 512,
111
+ 512,
112
+ 1500
113
+ ],
114
+ "tdnn_kernel": [
115
+ 5,
116
+ 3,
117
+ 3,
118
+ 1,
119
+ 1
120
+ ],
121
+ "transformers_version": "5.2.0",
122
+ "use_weighted_layer_sum": false,
123
+ "vocab_size": 32,
124
+ "xvector_output_dim": 512
125
+ }
label_map.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "label2id": {
3
+ "fake": 1,
4
+ "real": 0
5
+ },
6
+ "id2label": {
7
+ "1": "fake",
8
+ "0": "real"
9
+ }
10
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e6bc501e391e78a52f5ce76f34c726c6b30fe225c0d24a37a55daf284b5b70ec
3
+ size 378302312
preprocessor_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0.0,
7
+ "return_attention_mask": false,
8
+ "sampling_rate": 16000
9
+ }