Spaces:
Runtime error
Runtime error
ORI-Muchim
commited on
Commit
•
52542e1
1
Parent(s):
e882986
Upload 3 files
Browse files- .gitattributes +1 -0
- saved_model/config.json +55 -0
- saved_model/cover.png +0 -0
- saved_model/model.pth +3 -0
.gitattributes
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
saved_model/model.pth filter=lfs diff=lfs merge=lfs -text
|
saved_model/config.json
ADDED
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"train": {
|
3 |
+
"log_interval": 1000,
|
4 |
+
"eval_interval": 1000,
|
5 |
+
"seed": 1234,
|
6 |
+
"epochs": 10000,
|
7 |
+
"learning_rate": 2e-4,
|
8 |
+
"betas": [0.8, 0.99],
|
9 |
+
"eps": 1e-9,
|
10 |
+
"batch_size": 32,
|
11 |
+
"fp16_run": true,
|
12 |
+
"lr_decay": 0.999875,
|
13 |
+
"segment_size": 8192,
|
14 |
+
"init_lr_ratio": 1,
|
15 |
+
"warmup_epochs": 0,
|
16 |
+
"c_mel": 45,
|
17 |
+
"c_kl": 1.0
|
18 |
+
},
|
19 |
+
"data": {
|
20 |
+
"training_files":"../mdud_datasets/marin_train.txt.cleaned",
|
21 |
+
"validation_files":"../mdud_datasets/marin_val.txt.cleaned",
|
22 |
+
"text_cleaners":["japanese_cleaners2"],
|
23 |
+
"max_wav_value": 32768.0,
|
24 |
+
"sampling_rate": 44100,
|
25 |
+
"filter_length": 1024,
|
26 |
+
"hop_length": 256,
|
27 |
+
"win_length": 1024,
|
28 |
+
"n_mel_channels": 80,
|
29 |
+
"mel_fmin": 0.0,
|
30 |
+
"mel_fmax": null,
|
31 |
+
"add_blank": true,
|
32 |
+
"n_speakers": 0,
|
33 |
+
"cleaned_text": true
|
34 |
+
},
|
35 |
+
"model": {
|
36 |
+
"inter_channels": 192,
|
37 |
+
"hidden_channels": 192,
|
38 |
+
"filter_channels": 768,
|
39 |
+
"n_heads": 2,
|
40 |
+
"n_layers": 6,
|
41 |
+
"kernel_size": 3,
|
42 |
+
"p_dropout": 0.1,
|
43 |
+
"resblock": "1",
|
44 |
+
"resblock_kernel_sizes": [3,7,11],
|
45 |
+
"resblock_dilation_sizes": [[1,3,5], [1,3,5], [1,3,5]],
|
46 |
+
"upsample_rates": [8,8,2,2],
|
47 |
+
"upsample_initial_channel": 512,
|
48 |
+
"upsample_kernel_sizes": [16,16,4,4],
|
49 |
+
"n_layers_q": 3,
|
50 |
+
"use_spectral_norm": false,
|
51 |
+
"gin_channels": 256
|
52 |
+
},
|
53 |
+
"speakers": ["marin"],
|
54 |
+
"symbols": ["_", ",", ".", "!", "?", "-", "~", "\u2026", "A", "E", "I", "N", "O", "Q", "U", "a", "b", "d", "e", "f", "g", "h", "i", "j", "k", "m", "n", "o", "p", "r", "s", "t", "u", "v", "w", "y", "z", "\u0283", "\u02a7", "\u02a6", "\u2193", "\u2191", " "]
|
55 |
+
}
|
saved_model/cover.png
ADDED
saved_model/model.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1812823ef6cd01d98451103bc24d7f8da20f724bd9919f891d5e79bea2b79e90
|
3 |
+
size 449778950
|