Hugging Face's logo

Modularcomputing
/
dynamic-asr

Automatic Speech Recognition
ONNX
fairseq2
asr
speech-recognition
code-switching
nigerian-languages
igbo
yoruba
hausa
nigerian-pidgin
ctc
omnilingual-asr
streaming
Model card Files
xet
 Community
dynamic-asr
File size: 2,094 Bytes
e4c97d2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
{
 "base_model": "omniASR_CTC_1B_v2 (facebook omnilingual-asr, Apache-2.0)",
 "model_card": "omniASR_CTC_1B_v2",
 "architecture": "wav2vec2 CTC 1B (unchanged)",
 "vocab_size": 10288,
 "tokenizer": "omniASR_tokenizer_written_v2",
 "checkpoint_format": "torch.save dict {model: state_dict (bf16), step, dev, card}",
 "sample_rate": 16000,
 "decoding": "greedy CTC, Latin-script logit mask, no language hint",
 "languages": [
  "en",
  "pcm",
  "ig",
  "yo",
  "ha"
 ],
 "license": "cc-by-sa-4.0",
 "training": {
  "steps": 4000,
  "batch_audio_sec": 160,
  "max_clip_sec": 18,
  "lr": 2e-05,
  "warmup": 300,
  "schedule": "linear decay",
  "optimizer": "AdamW b(0.9,0.98) wd 0.01",
  "precision": "bf16 autocast",
  "frozen": "CNN feature encoder",
  "seed": 1,
  "real_aug_p": 0.15,
  "synthetic_codeswitch_fraction": 0.12,
  "data": "same licensed real speech + synthetic code-switched audio (Dynamic Accent Flash-DPO TTS), 12% of sampled audio",
  "synth_aug_p": 0.8
 },
 "name": "dynamic-asr/codeswitch-experimental",
 "status": "experimental",
 "file": "model.pt",
 "scores": {
  "duplex_eng": {
   "n": 29,
   "wer": 0.0922,
   "cer": 0.0241,
   "cer_nodiac": 0.0241
  },
  "duplex_pcm": {
   "n": 40,
   "wer": 0.1927,
   "cer": 0.0761,
   "cer_nodiac": 0.0759
  },
  "duplex_ibo": {
   "n": 40,
   "wer": 0.3207,
   "cer": 0.095,
   "cer_nodiac": 0.0754
  },
  "duplex_yor": {
   "n": 40,
   "wer": 0.3532,
   "cer": 0.1182,
   "cer_nodiac": 0.0782
  },
  "duplex_hau": {
   "n": 40,
   "wer": 0.252,
   "cer": 0.0808,
   "cer_nodiac": 0.0808
  },
  "duplex_cs": {
   "n": 40,
   "wer": 0.2157,
   "cer": 0.0718,
   "cer_nodiac": 0.0648
  },
  "realcs_all": {
   "n": 94,
   "wer": 0.438,
   "cer": 0.1492,
   "cer_nodiac": 0.1368
  },
  "realcs_natural_conversational": {
   "n": 36,
   "wer": 0.6103,
   "cer": 0.2472,
   "cer_nodiac": 0.2339
  },
  "realcs_insertional": {
   "n": 58,
   "wer": 0.3727,
   "cer": 0.1186,
   "cer_nodiac": 0.1065
  },
  "realcs_eng_word_recall": 0.3805,
  "synthtest": {
   "n": 109,
   "wer": 0.1261,
   "cer": 0.0385,
   "cer_nodiac": 0.0349
  }
 }
}