hyunwoo3235
commited on
Commit
•
d856028
1
Parent(s):
0a741dc
Upload 6 files
Browse files- classifier.ckpt +3 -0
- config.json +3 -0
- embedding_model.ckpt +3 -0
- hyperparams.yaml +63 -0
- label_encoder.txt +0 -0
- mean_var_norm_emb.ckpt +3 -0
classifier.ckpt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5f7ea89eb44b3a3e8e3b957b8a473f9b024c2686e95170963223d2182ff1c076
|
3 |
+
size 13207403
|
config.json
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"speechbrain_interface": "EncoderClassifier"
|
3 |
+
}
|
embedding_model.ckpt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:17890d7a013d36d71071599cb0ac720b673b5b16863c8e1ad769c3ce525b57d3
|
3 |
+
size 16887086
|
hyperparams.yaml
ADDED
@@ -0,0 +1,63 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# ############################################################################
|
2 |
+
# Model: xvector for Speaker verification
|
3 |
+
# ############################################################################
|
4 |
+
|
5 |
+
# Feature parameters
|
6 |
+
n_mels: 24
|
7 |
+
|
8 |
+
# Pretrain folder (HuggingFace)
|
9 |
+
pretrained_path: team-lucid/spkrec-xvect-korean
|
10 |
+
|
11 |
+
# Output parameters
|
12 |
+
out_n_neurons: 5914
|
13 |
+
|
14 |
+
|
15 |
+
# Model params
|
16 |
+
compute_features: !new:speechbrain.lobes.features.Fbank
|
17 |
+
n_mels: !ref <n_mels>
|
18 |
+
|
19 |
+
mean_var_norm: !new:speechbrain.processing.features.InputNormalization
|
20 |
+
norm_type: sentence
|
21 |
+
std_norm: False
|
22 |
+
|
23 |
+
embedding_model: !new:speechbrain.lobes.models.Xvector.Xvector
|
24 |
+
in_channels: !ref <n_mels>
|
25 |
+
activation: !name:torch.nn.LeakyReLU
|
26 |
+
tdnn_blocks: 5
|
27 |
+
tdnn_channels: [512, 512, 512, 512, 1500]
|
28 |
+
tdnn_kernel_sizes: [5, 3, 3, 1, 1]
|
29 |
+
tdnn_dilations: [1, 2, 3, 1, 1]
|
30 |
+
lin_neurons: 512
|
31 |
+
|
32 |
+
classifier: !new:speechbrain.lobes.models.Xvector.Classifier
|
33 |
+
input_shape: [null, null, 512]
|
34 |
+
activation: !name:torch.nn.LeakyReLU
|
35 |
+
lin_blocks: 1
|
36 |
+
lin_neurons: 512
|
37 |
+
out_neurons: !ref <out_n_neurons>
|
38 |
+
|
39 |
+
mean_var_norm_emb: !new:speechbrain.processing.features.InputNormalization
|
40 |
+
norm_type: global
|
41 |
+
std_norm: False
|
42 |
+
|
43 |
+
modules:
|
44 |
+
compute_features: !ref <compute_features>
|
45 |
+
mean_var_norm: !ref <mean_var_norm>
|
46 |
+
embedding_model: !ref <embedding_model>
|
47 |
+
mean_var_norm_emb: !ref <mean_var_norm_emb>
|
48 |
+
classifier: !ref <classifier>
|
49 |
+
|
50 |
+
label_encoder: !new:speechbrain.dataio.encoder.CategoricalEncoder
|
51 |
+
|
52 |
+
|
53 |
+
pretrainer: !new:speechbrain.utils.parameter_transfer.Pretrainer
|
54 |
+
loadables:
|
55 |
+
embedding_model: !ref <embedding_model>
|
56 |
+
mean_var_norm_emb: !ref <mean_var_norm_emb>
|
57 |
+
classifier: !ref <classifier>
|
58 |
+
label_encoder: !ref <label_encoder>
|
59 |
+
paths:
|
60 |
+
embedding_model: !ref <pretrained_path>/embedding_model.ckpt
|
61 |
+
mean_var_norm_emb: !ref <pretrained_path>/mean_var_norm_emb.ckpt
|
62 |
+
classifier: !ref <pretrained_path>/classifier.ckpt
|
63 |
+
label_encoder: !ref <pretrained_path>/label_encoder.txt
|
label_encoder.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
mean_var_norm_emb.ckpt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:26438e2995f675d14eb728a7e794794526db4c4d4c3fa06f7ba48e2e211a3341
|
3 |
+
size 1167
|