Upload 2 files
Browse files- config.yaml +23 -0
- pytorch_model.bin +3 -0
config.yaml
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
feature_extractor:
|
| 2 |
+
class_path: vocos.feature_extractors.EncodecFeatures
|
| 3 |
+
init_args:
|
| 4 |
+
encodec_model: encodec_24khz
|
| 5 |
+
bandwidths: [1.5, 3.0, 6.0, 12.0]
|
| 6 |
+
train_codebooks: false
|
| 7 |
+
|
| 8 |
+
backbone:
|
| 9 |
+
class_path: vocos.models.VocosBackbone
|
| 10 |
+
init_args:
|
| 11 |
+
input_channels: 128
|
| 12 |
+
dim: 384
|
| 13 |
+
intermediate_dim: 1152
|
| 14 |
+
num_layers: 8
|
| 15 |
+
adanorm_num_embeddings: 4 # len(bandwidths)
|
| 16 |
+
|
| 17 |
+
head:
|
| 18 |
+
class_path: vocos.heads.ISTFTHead
|
| 19 |
+
init_args:
|
| 20 |
+
dim: 384
|
| 21 |
+
n_fft: 1280
|
| 22 |
+
hop_length: 320
|
| 23 |
+
padding: same
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7e95bb260b74a1bfc43c52d355831c951acb81c8960e9c62b79bd2b3ab1e3a90
|
| 3 |
+
size 40356708
|