Hugging Face's logo

wiftwift
/
accent-scorerer

Audio Classification
English
accentscore
audio
accent-classification
pronunciation-assessment
wavlm
research
Model card Files
xet
 Community
accent-scorerer
File size: 2,682 Bytes
968b170
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
{
  "schema_version": "2.0",
  "model_version": "0.2.1",
  "model_type": "wavlm_layer_mix",
  "checkpoint_sha256": "2c516f95574aa40f800e6900f0db5a6f7804e80f9366a50b506c22e841c33e40",
  "architecture": {
    "encoder_name": "microsoft/wavlm-base-plus",
    "encoder_revision": "4c66d4806a428f2e922ccfa1a962776e232d487b",
    "num_layers": 13,
    "hidden_size": 768,
    "projection_size": 128,
    "dropout": 0.15,
    "target_specific_layer_mixture": true
  },
  "target_names": [
    "accuracy",
    "fluency",
    "prosody",
    "completeness",
    "overall"
  ],
  "accent": {
    "checkpoint": "accent.pt",
    "model_version": "0.2.1-aesrc-saa",
    "labels": [
      "american",
      "british",
      "indian",
      "russian"
    ],
    "selected_wavlm_layer": 7,
    "classifier": "L2 logistic regression",
    "regularization_c": 0.01,
    "speech_accent_archive_domain_mass_relative_to_aesrc": 2.0,
    "temperature": 1.0126516733940019,
    "unknown_probability_threshold": 0.4,
    "unknown_threshold_policy": "fixed at 0.4 before held-out evaluation",
    "validation_macro_f1": {
      "aesrc": 0.8919,
      "speech_accent_archive": 0.9644,
      "cross_domain_mean": 0.9281
    },
    "speaker_disjoint_test_macro_f1": {
      "aesrc": 0.9047,
      "speech_accent_archive": 0.895,
      "cross_domain_mean": 0.8998
    },
    "speaker_disjoint_test_accuracy": {
      "aesrc": 0.9055,
      "speech_accent_archive": 0.939
    },
    "training_corpora": [
      "AESRC2020",
      "Speech Accent Archive"
    ],
    "training_corpus_revision": "4a80d8388f06368a0fa2a325770bec3492cabd3d",
    "speech_accent_archive_metadata_sha256": "979afd5a7b0b53fa5d8e8a9b8587a1fbc2de8567d4c6b6c7fbaead28d55cf33b",
    "checkpoint_sha256": "03fe84174ceacdae4dc487e622f26814da00704a4d0a5747a836c52b391263fe",
    "license_status": "research use only; verify AESRC and Speech Accent Archive training and derivative-work rights before redistribution"
  },
  "american_similarity": {
    "checkpoint": "american_similarity.pt",
    "model_version": "0.2.0-aesrc",
    "definition": "100 * calibrated P(American vs non-American)",
    "selected_wavlm_layer": 6,
    "calibration": "temperature scaling on speaker-disjoint validation",
    "validation_roc_auc": 0.9771,
    "speaker_disjoint_test_roc_auc": 0.9607,
    "checkpoint_sha256": "ebd61bdc6e680ee58fff1d85d2ff3b78c42ea06635c5d4fe6bdb321d4b1222f1"
  },
  "recommended_mps_dtype": "float16",
  "dtype_qualification": {
    "recommended_mps_dtype": "float16",
    "criteria": {
      "finite_predictions": true,
      "max_score_delta_points": 1.0,
      "minimum_p50_speedup_fraction": 0.05
    },
    "qualified": true
  }
}