Upload 5 files
Browse files- .gitattributes +1 -0
- ATRI_G_40000.pth +3 -0
- ATRI_diffmodel_100000.pt +3 -0
- ATRI_diffusion.yaml +49 -0
- ATRI_kmeans_10000.pt +3 -0
- demo/fullsong_12.0key_ATRI_0.1_sovdiff.wav +3 -0
.gitattributes
CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
demo/fullsong_12.0key_ATRI_0.1_sovdiff.wav filter=lfs diff=lfs merge=lfs -text
|
ATRI_G_40000.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6f291ed8b74bfeb2ebd162e91e9f8f1300a7ae163a7ad24da6a6a7dcd2b05b0a
|
3 |
+
size 627905309
|
ATRI_diffmodel_100000.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d845594dcf12815f39601e465f8384ac5a0e2a0dacfe381ad53300a45f27f622
|
3 |
+
size 220893960
|
ATRI_diffusion.yaml
ADDED
@@ -0,0 +1,49 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
data:
|
2 |
+
block_size: 512
|
3 |
+
cnhubertsoft_gate: 10
|
4 |
+
duration: 2
|
5 |
+
encoder: vec768l12
|
6 |
+
encoder_hop_size: 320
|
7 |
+
encoder_out_channels: 768
|
8 |
+
encoder_sample_rate: 16000
|
9 |
+
extensions:
|
10 |
+
- wav
|
11 |
+
sampling_rate: 44100
|
12 |
+
training_files: filelists/train.txt
|
13 |
+
unit_interpolate_mode: nearest
|
14 |
+
validation_files: filelists/val.txt
|
15 |
+
device: cuda
|
16 |
+
env:
|
17 |
+
expdir: logs/44k/diffusion
|
18 |
+
gpu_id: 0
|
19 |
+
infer:
|
20 |
+
method: dpm-solver
|
21 |
+
speedup: 10
|
22 |
+
model:
|
23 |
+
n_chans: 512
|
24 |
+
n_hidden: 256
|
25 |
+
n_layers: 20
|
26 |
+
n_spk: 1
|
27 |
+
type: Diffusion
|
28 |
+
use_pitch_aug: true
|
29 |
+
spk:
|
30 |
+
ATRI: 0
|
31 |
+
train:
|
32 |
+
amp_dtype: fp16
|
33 |
+
batch_size: 48
|
34 |
+
cache_all_data: true
|
35 |
+
cache_device: cuda
|
36 |
+
cache_fp16: true
|
37 |
+
decay_step: 100000
|
38 |
+
epochs: 100000
|
39 |
+
gamma: 0.5
|
40 |
+
interval_force_save: 2000
|
41 |
+
interval_log: 10
|
42 |
+
interval_val: 2000
|
43 |
+
lr: 0.0002
|
44 |
+
num_workers: 0
|
45 |
+
save_opt: false
|
46 |
+
weight_decay: 0
|
47 |
+
vocoder:
|
48 |
+
ckpt: pretrain/nsf_hifigan/model
|
49 |
+
type: nsf-hifigan
|
ATRI_kmeans_10000.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e8b91b8003b3883f9dc2e525926bc6ad670c297d4126a4aa00410b612f56ec0b
|
3 |
+
size 31048057
|
demo/fullsong_12.0key_ATRI_0.1_sovdiff.wav
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9f4b00de99fe3813b92eb75fc01ccf5ef2bf7377846ebece219e1e0be2b69095
|
3 |
+
size 22855608
|