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
|
| # network architecture
| model: FsmnKWS
| model_conf:
| ctc_weight: 1.0
|
| # encoder related
| encoder: FSMN
| encoder_conf:
| input_dim: 360
| input_affine_dim: 280
| fsmn_layers: 4
| linear_dim: 280
| proj_dim: 200
| lorder: 10
| rorder: 2
| lstride: 1
| rstride: 1
| output_affine_dim: 400
| output_dim: 2602
| use_softmax: false
|
| frontend: WavFrontend
| frontend_conf:
| fs: 16000
| window: hamming
| n_mels: 40
| frame_length: 25
| frame_shift: 10
| lfr_m: 9
| lfr_n: 3
|
| specaug: SpecAugLFR
| specaug_conf:
| apply_time_warp: false
| time_warp_window: 5
| time_warp_mode: bicubic
| apply_freq_mask: true
| freq_mask_width_range:
| - 0
| - 30
| lfr_rate: 3
| num_freq_mask: 1
| apply_time_mask: true
| time_mask_width_range:
| - 0
| - 12
| num_time_mask: 1
|
| train_conf:
| accum_grad: 1
| grad_clip: 5
| max_epoch: 100
| keep_nbest_models: 10
| avg_nbest_model: 10
| avg_keep_nbest_models_type: loss
| validate_interval: 50000
| save_checkpoint_interval: 50000
| avg_checkpoint_interval: 1000
| log_interval: 50
|
| optim: adam
| optim_conf:
| lr: 0.0005
| scheduler: warmuplr
| scheduler_conf:
| warmup_steps: 10000
|
| dataset: AudioDataset
| dataset_conf:
| index_ds: IndexDSJsonl
| batch_sampler: EspnetStyleBatchSampler
| batch_type: length # example or length
| batch_size: 32000 # if batch_type is example, batch_size is the numbers of samples; if length, batch_size is source_token_len+target_token_len;
| max_token_length: 1600 # filter samples if source_token_len+target_token_len > max_token_length,
| buffer_size: 2048
| shuffle: true
| num_workers: 8
| preprocessor_speech: SpeechPreprocessSpeedPerturb
| preprocessor_speech_conf:
| speed_perturb: [0.9, 1.0, 1.1]
|
| tokenizer: CharTokenizer
| tokenizer_conf:
| unk_symbol: <unk>
| split_with_space: true
|
| ctc_conf:
| dropout_rate: 0.0
| ctc_type: builtin
| reduce: true
| ignore_nan_grad: true
| extra_linear: false
|
| normalize: null
|
|
