Upload 2 files
Browse files- config.yml +60 -0
- model.h5 +3 -0
config.yml
ADDED
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
allow_cache: true
|
2 |
+
batch_size: 16
|
3 |
+
config: /content/TensorFlowTTS/examples/fastspeech2/conf/fastspeech2.v1.yaml
|
4 |
+
dev_dir: /content/dump_synpaflex/valid/
|
5 |
+
energy_stat: /content/dump_synpaflex/stats_energy.npy
|
6 |
+
eval_interval_steps: 500
|
7 |
+
f0_stat: /content/dump_synpaflex/stats_f0.npy
|
8 |
+
fastspeech2_params:
|
9 |
+
attention_probs_dropout_prob: 0.1
|
10 |
+
decoder_attention_head_size: 192
|
11 |
+
decoder_hidden_act: mish
|
12 |
+
decoder_hidden_size: 384
|
13 |
+
decoder_intermediate_kernel_size: 3
|
14 |
+
decoder_intermediate_size: 1024
|
15 |
+
decoder_num_attention_heads: 2
|
16 |
+
decoder_num_hidden_layers: 4
|
17 |
+
encoder_attention_head_size: 192
|
18 |
+
encoder_hidden_act: mish
|
19 |
+
encoder_hidden_size: 384
|
20 |
+
encoder_intermediate_kernel_size: 3
|
21 |
+
encoder_intermediate_size: 1024
|
22 |
+
encoder_num_attention_heads: 2
|
23 |
+
encoder_num_hidden_layers: 4
|
24 |
+
hidden_dropout_prob: 0.2
|
25 |
+
initializer_range: 0.02
|
26 |
+
max_position_embeddings: 2048
|
27 |
+
n_speakers: 1
|
28 |
+
num_mels: 80
|
29 |
+
output_attentions: false
|
30 |
+
output_hidden_states: false
|
31 |
+
variant_prediction_num_conv_layers: 2
|
32 |
+
variant_predictor_dropout_rate: 0.5
|
33 |
+
variant_predictor_filter: 256
|
34 |
+
variant_predictor_kernel_size: 3
|
35 |
+
format: npy
|
36 |
+
gradient_accumulation_steps: 1
|
37 |
+
hop_size: 256
|
38 |
+
is_shuffle: true
|
39 |
+
log_interval_steps: 200
|
40 |
+
mel_length_threshold: 32
|
41 |
+
mixed_precision: true
|
42 |
+
model_type: fastspeech2
|
43 |
+
num_save_intermediate_results: 1
|
44 |
+
optimizer_params:
|
45 |
+
decay_steps: 150000
|
46 |
+
end_learning_rate: 5.0e-05
|
47 |
+
initial_learning_rate: 0.001
|
48 |
+
warmup_proportion: 0.02
|
49 |
+
weight_decay: 0.001
|
50 |
+
outdir: /content/drive/MyDrive/exp/train.fastspeech2.v1/
|
51 |
+
pretrained: ''
|
52 |
+
remove_short_samples: true
|
53 |
+
resume: /content/drive/MyDrive/exp/train.fastspeech2.v1/checkpoints/ckpt-1.index
|
54 |
+
save_interval_steps: 5000
|
55 |
+
train_dir: /content/dump_synpaflex/train/
|
56 |
+
train_max_steps: 200000
|
57 |
+
use_norm: true
|
58 |
+
var_train_expr: null
|
59 |
+
verbose: 1
|
60 |
+
version: '0.0'
|
model.h5
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d68714458e51a3a4f91f602570c25218d45bb078f5a18ee2e525343350b2acd6
|
3 |
+
size 124796368
|