ahnafsamin
commited on
Commit
•
a09f3e8
1
Parent(s):
3234b81
Upload config.yaml
Browse files- config.yaml +257 -0
config.yaml
ADDED
@@ -0,0 +1,257 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
config: conf/tuning/train_fastspeech2.yaml
|
2 |
+
print_config: false
|
3 |
+
log_level: INFO
|
4 |
+
dry_run: false
|
5 |
+
iterator_type: sequence
|
6 |
+
output_dir: exp/tts_train_fastspeech2_raw_char_tacotron
|
7 |
+
ngpu: 1
|
8 |
+
seed: 0
|
9 |
+
num_workers: 1
|
10 |
+
num_att_plot: 3
|
11 |
+
dist_backend: nccl
|
12 |
+
dist_init_method: env://
|
13 |
+
dist_world_size: null
|
14 |
+
dist_rank: null
|
15 |
+
local_rank: 0
|
16 |
+
dist_master_addr: null
|
17 |
+
dist_master_port: null
|
18 |
+
dist_launcher: null
|
19 |
+
multiprocessing_distributed: false
|
20 |
+
unused_parameters: false
|
21 |
+
sharded_ddp: false
|
22 |
+
cudnn_enabled: true
|
23 |
+
cudnn_benchmark: false
|
24 |
+
cudnn_deterministic: true
|
25 |
+
collect_stats: false
|
26 |
+
write_collected_feats: false
|
27 |
+
max_epoch: 1000
|
28 |
+
patience: null
|
29 |
+
val_scheduler_criterion:
|
30 |
+
- valid
|
31 |
+
- loss
|
32 |
+
early_stopping_criterion:
|
33 |
+
- valid
|
34 |
+
- loss
|
35 |
+
- min
|
36 |
+
best_model_criterion:
|
37 |
+
- - valid
|
38 |
+
- loss
|
39 |
+
- min
|
40 |
+
- - train
|
41 |
+
- loss
|
42 |
+
- min
|
43 |
+
keep_nbest_models: 5
|
44 |
+
nbest_averaging_interval: 0
|
45 |
+
grad_clip: 1.0
|
46 |
+
grad_clip_type: 2.0
|
47 |
+
grad_noise: false
|
48 |
+
accum_grad: 8
|
49 |
+
no_forward_run: false
|
50 |
+
resume: true
|
51 |
+
train_dtype: float32
|
52 |
+
use_amp: false
|
53 |
+
log_interval: null
|
54 |
+
use_matplotlib: true
|
55 |
+
use_tensorboard: true
|
56 |
+
use_wandb: false
|
57 |
+
wandb_project: null
|
58 |
+
wandb_id: null
|
59 |
+
wandb_entity: null
|
60 |
+
wandb_name: null
|
61 |
+
wandb_model_log_interval: -1
|
62 |
+
detect_anomaly: false
|
63 |
+
pretrain_path: null
|
64 |
+
init_param: []
|
65 |
+
ignore_init_mismatch: false
|
66 |
+
freeze_param: []
|
67 |
+
num_iters_per_epoch: 800
|
68 |
+
batch_size: 20
|
69 |
+
valid_batch_size: null
|
70 |
+
batch_bins: 3000000
|
71 |
+
valid_batch_bins: null
|
72 |
+
train_shape_file:
|
73 |
+
- exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/train/text_shape.char
|
74 |
+
- exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/train/speech_shape
|
75 |
+
valid_shape_file:
|
76 |
+
- exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/valid/text_shape.char
|
77 |
+
- exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/valid/speech_shape
|
78 |
+
batch_type: numel
|
79 |
+
valid_batch_type: null
|
80 |
+
fold_length:
|
81 |
+
- 150
|
82 |
+
- 204800
|
83 |
+
sort_in_batch: descending
|
84 |
+
sort_batch: descending
|
85 |
+
multiple_iterator: false
|
86 |
+
chunk_length: 500
|
87 |
+
chunk_shift_ratio: 0.5
|
88 |
+
num_cache_chunks: 1024
|
89 |
+
train_data_path_and_name_and_type:
|
90 |
+
- - dump/raw/tr_no_dev/text
|
91 |
+
- text
|
92 |
+
- text
|
93 |
+
- - exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/tr_no_dev/durations
|
94 |
+
- durations
|
95 |
+
- text_int
|
96 |
+
- - dump/raw/tr_no_dev/wav.scp
|
97 |
+
- speech
|
98 |
+
- sound
|
99 |
+
- - exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/train/collect_feats/pitch.scp
|
100 |
+
- pitch
|
101 |
+
- npy
|
102 |
+
- - exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/train/collect_feats/energy.scp
|
103 |
+
- energy
|
104 |
+
- npy
|
105 |
+
valid_data_path_and_name_and_type:
|
106 |
+
- - dump/raw/dev/text
|
107 |
+
- text
|
108 |
+
- text
|
109 |
+
- - exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/dev/durations
|
110 |
+
- durations
|
111 |
+
- text_int
|
112 |
+
- - dump/raw/dev/wav.scp
|
113 |
+
- speech
|
114 |
+
- sound
|
115 |
+
- - exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/valid/collect_feats/pitch.scp
|
116 |
+
- pitch
|
117 |
+
- npy
|
118 |
+
- - exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/valid/collect_feats/energy.scp
|
119 |
+
- energy
|
120 |
+
- npy
|
121 |
+
allow_variable_data_keys: false
|
122 |
+
max_cache_size: 0.0
|
123 |
+
max_cache_fd: 32
|
124 |
+
valid_max_cache_size: null
|
125 |
+
optim: adam
|
126 |
+
optim_conf:
|
127 |
+
lr: 1.0
|
128 |
+
scheduler: noamlr
|
129 |
+
scheduler_conf:
|
130 |
+
model_size: 384
|
131 |
+
warmup_steps: 4000
|
132 |
+
token_list:
|
133 |
+
- <blank>
|
134 |
+
- <unk>
|
135 |
+
- <space>
|
136 |
+
- E
|
137 |
+
- N
|
138 |
+
- A
|
139 |
+
- O
|
140 |
+
- T
|
141 |
+
- I
|
142 |
+
- R
|
143 |
+
- D
|
144 |
+
- L
|
145 |
+
- S
|
146 |
+
- K
|
147 |
+
- M
|
148 |
+
- U
|
149 |
+
- G
|
150 |
+
- H
|
151 |
+
- W
|
152 |
+
- V
|
153 |
+
- Z
|
154 |
+
- P
|
155 |
+
- B
|
156 |
+
- .
|
157 |
+
- J
|
158 |
+
- C
|
159 |
+
- ','
|
160 |
+
- F
|
161 |
+
- ''''
|
162 |
+
- '?'
|
163 |
+
- '!'
|
164 |
+
- Y
|
165 |
+
- X
|
166 |
+
- '`'
|
167 |
+
- <sos/eos>
|
168 |
+
odim: null
|
169 |
+
model_conf: {}
|
170 |
+
use_preprocessor: true
|
171 |
+
token_type: char
|
172 |
+
bpemodel: null
|
173 |
+
non_linguistic_symbols: null
|
174 |
+
cleaner: tacotron
|
175 |
+
g2p: g2p_en
|
176 |
+
feats_extract: fbank
|
177 |
+
feats_extract_conf:
|
178 |
+
n_fft: 1024
|
179 |
+
hop_length: 256
|
180 |
+
win_length: null
|
181 |
+
fs: 22050
|
182 |
+
fmin: 80
|
183 |
+
fmax: 7600
|
184 |
+
n_mels: 80
|
185 |
+
normalize: global_mvn
|
186 |
+
normalize_conf:
|
187 |
+
stats_file: exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/train/feats_stats.npz
|
188 |
+
tts: fastspeech2
|
189 |
+
tts_conf:
|
190 |
+
adim: 384
|
191 |
+
aheads: 2
|
192 |
+
elayers: 4
|
193 |
+
eunits: 1536
|
194 |
+
dlayers: 4
|
195 |
+
dunits: 1536
|
196 |
+
positionwise_layer_type: conv1d
|
197 |
+
positionwise_conv_kernel_size: 3
|
198 |
+
duration_predictor_layers: 2
|
199 |
+
duration_predictor_chans: 256
|
200 |
+
duration_predictor_kernel_size: 3
|
201 |
+
postnet_layers: 5
|
202 |
+
postnet_filts: 5
|
203 |
+
postnet_chans: 256
|
204 |
+
use_masking: true
|
205 |
+
use_scaled_pos_enc: true
|
206 |
+
encoder_normalize_before: true
|
207 |
+
decoder_normalize_before: true
|
208 |
+
reduction_factor: 1
|
209 |
+
init_type: xavier_uniform
|
210 |
+
init_enc_alpha: 1.0
|
211 |
+
init_dec_alpha: 1.0
|
212 |
+
transformer_enc_dropout_rate: 0.2
|
213 |
+
transformer_enc_positional_dropout_rate: 0.2
|
214 |
+
transformer_enc_attn_dropout_rate: 0.2
|
215 |
+
transformer_dec_dropout_rate: 0.2
|
216 |
+
transformer_dec_positional_dropout_rate: 0.2
|
217 |
+
transformer_dec_attn_dropout_rate: 0.2
|
218 |
+
pitch_predictor_layers: 5
|
219 |
+
pitch_predictor_chans: 256
|
220 |
+
pitch_predictor_kernel_size: 5
|
221 |
+
pitch_predictor_dropout: 0.5
|
222 |
+
pitch_embed_kernel_size: 1
|
223 |
+
pitch_embed_dropout: 0.0
|
224 |
+
stop_gradient_from_pitch_predictor: true
|
225 |
+
energy_predictor_layers: 2
|
226 |
+
energy_predictor_chans: 256
|
227 |
+
energy_predictor_kernel_size: 3
|
228 |
+
energy_predictor_dropout: 0.5
|
229 |
+
energy_embed_kernel_size: 1
|
230 |
+
energy_embed_dropout: 0.0
|
231 |
+
stop_gradient_from_energy_predictor: false
|
232 |
+
pitch_extract: dio
|
233 |
+
pitch_extract_conf:
|
234 |
+
fs: 22050
|
235 |
+
n_fft: 1024
|
236 |
+
hop_length: 256
|
237 |
+
f0max: 400
|
238 |
+
f0min: 80
|
239 |
+
reduction_factor: 1
|
240 |
+
pitch_normalize: global_mvn
|
241 |
+
pitch_normalize_conf:
|
242 |
+
stats_file: exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/train/pitch_stats.npz
|
243 |
+
energy_extract: energy
|
244 |
+
energy_extract_conf:
|
245 |
+
fs: 22050
|
246 |
+
n_fft: 1024
|
247 |
+
hop_length: 256
|
248 |
+
win_length: null
|
249 |
+
reduction_factor: 1
|
250 |
+
energy_normalize: global_mvn
|
251 |
+
energy_normalize_conf:
|
252 |
+
stats_file: exp/tts_train_raw_char_tacotron/decode_use_teacher_forcingtrue_train.loss.ave/stats/train/energy_stats.npz
|
253 |
+
required:
|
254 |
+
- output_dir
|
255 |
+
- token_list
|
256 |
+
version: 0.10.6
|
257 |
+
distributed: false
|