Hugging Face's logo

espnet
/
kan-bayashi_jsut_transformer

Text-to-Speech
ESPnet
Japanese
audio
Model card Files Community
kan-bayashi_jsut_transformer
File size: 4,145 Bytes 
1328b49
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
 
config: conf/tuning/train_transformer.yaml
print_config: false
log_level: INFO
dry_run: false
iterator_type: sequence
output_dir: exp/tts_train_transformer_raw_phn_jaconv_pyopenjtalk
ngpu: 1
seed: 0
num_workers: 1
num_att_plot: 3
dist_backend: nccl
dist_init_method: env://
dist_world_size: 4
dist_rank: 0
local_rank: 0
dist_master_addr: localhost
dist_master_port: 58206
dist_launcher: null
multiprocessing_distributed: true
cudnn_enabled: true
cudnn_benchmark: false
cudnn_deterministic: true
collect_stats: false
write_collected_feats: false
max_epoch: 200
patience: null
val_scheduler_criterion:
- valid
- loss
early_stopping_criterion:
- valid
- loss
- min
best_model_criterion:
-   - valid
    - loss
    - min
-   - train
    - loss
    - min
keep_nbest_models: 5
grad_clip: 1.0
grad_clip_type: 2.0
grad_noise: false
accum_grad: 2
no_forward_run: false
resume: true
train_dtype: float32
use_amp: false
log_interval: null
pretrain_path: []
pretrain_key: []
num_iters_per_epoch: 1000
batch_size: 20
valid_batch_size: null
batch_bins: 9000000
valid_batch_bins: null
train_shape_file:
- exp/tts_stats_raw_phn_jaconv_pyopenjtalk/train/text_shape.phn
- exp/tts_stats_raw_phn_jaconv_pyopenjtalk/train/speech_shape
valid_shape_file:
- exp/tts_stats_raw_phn_jaconv_pyopenjtalk/valid/text_shape.phn
- exp/tts_stats_raw_phn_jaconv_pyopenjtalk/valid/speech_shape
batch_type: numel
valid_batch_type: null
fold_length:
- 150
- 240000
sort_in_batch: descending
sort_batch: descending
multiple_iterator: false
chunk_length: 500
chunk_shift_ratio: 0.5
num_cache_chunks: 1024
train_data_path_and_name_and_type:
-   - dump/raw/tr_no_dev/text
    - text
    - text
-   - dump/raw/tr_no_dev/wav.scp
    - speech
    - sound
valid_data_path_and_name_and_type:
-   - dump/raw/dev/text
    - text
    - text
-   - dump/raw/dev/wav.scp
    - speech
    - sound
allow_variable_data_keys: false
max_cache_size: 0.0
valid_max_cache_size: null
optim: adam
optim_conf:
    lr: 1.0
scheduler: noamlr
scheduler_conf:
    model_size: 512
    warmup_steps: 8000
token_list:
- <blank>
- <unk>
- a
- o
- i
- u
- e
- k
- n
- t
- r
- s
- N
- m
- pau
- sh
- d
- g
- w
- U
- I
- cl
- h
- y
- b
- j
- ts
- ch
- z
- p
- f
- ky
- ry
- gy
- hy
- ny
- by
- my
- py
- v
- dy
- ty
- <sos/eos>
odim: null
model_conf: {}
use_preprocessor: true
token_type: phn
bpemodel: null
non_linguistic_symbols: null
cleaner: jaconv
g2p: pyopenjtalk
feats_extract: fbank
feats_extract_conf:
    fs: 24000
    fmin: 80
    fmax: 7600
    n_mels: 80
    hop_length: 300
    n_fft: 2048
    win_length: 1200
normalize: global_mvn
normalize_conf:
    stats_file: exp/tts_stats_raw_phn_jaconv_pyopenjtalk/train/feats_stats.npz
tts: transformer
tts_conf:
    embed_dim: 0
    eprenet_conv_layers: 0
    eprenet_conv_filts: 0
    eprenet_conv_chans: 0
    dprenet_layers: 2
    dprenet_units: 256
    adim: 512
    aheads: 8
    elayers: 6
    eunits: 1024
    dlayers: 6
    dunits: 1024
    positionwise_layer_type: conv1d
    positionwise_conv_kernel_size: 1
    postnet_layers: 5
    postnet_filts: 5
    postnet_chans: 256
    use_masking: true
    bce_pos_weight: 5.0
    use_scaled_pos_enc: true
    encoder_normalize_before: false
    decoder_normalize_before: false
    reduction_factor: 1
    init_type: xavier_uniform
    init_enc_alpha: 1.0
    init_dec_alpha: 1.0
    eprenet_dropout_rate: 0.0
    dprenet_dropout_rate: 0.5
    postnet_dropout_rate: 0.5
    transformer_enc_dropout_rate: 0.1
    transformer_enc_positional_dropout_rate: 0.1
    transformer_enc_attn_dropout_rate: 0.1
    transformer_dec_dropout_rate: 0.1
    transformer_dec_positional_dropout_rate: 0.1
    transformer_dec_attn_dropout_rate: 0.1
    transformer_enc_dec_attn_dropout_rate: 0.1
    use_guided_attn_loss: true
    num_heads_applied_guided_attn: 2
    num_layers_applied_guided_attn: 2
    modules_applied_guided_attn:
    - encoder-decoder
    guided_attn_loss_lambda: 10.0
pitch_extract: null
pitch_extract_conf: {}
pitch_normalize: null
pitch_normalize_conf: {}
energy_extract: null
energy_extract_conf: {}
energy_normalize: null
energy_normalize_conf: {}
required:
- output_dir
- token_list
distributed: true