Hugging Face's logo

danhtran2mind
/
project-Make-An-Audio-LDM

Model card Files
xet
Community
danhtran2mind commited on
Commit
8197e2f
·
verified ·
1 Parent(s): bc1626c

Add files using upload-large-folder tool

Browse files
Files changed (9) hide show
  1. .DS_Store +0 -0
  2. CLAP/CLAP_weights_2022.pth +3 -0
  3. CLAP/CLAP_weights_2023.pth +3 -0
  4. CLAP/clapcap_weights_2023.pth +3 -0
  5. CLAP/config.yml +26 -0
  6. bigvgan/args.yml +63 -0
  7. bigvgan/best_netG.pt +3 -0
  8. maa1_caps.ckpt +3 -0
  9. maa1_full.ckpt +3 -0
.DS_Store ADDED
Binary file (6.15 kB). View file
 
CLAP/CLAP_weights_2022.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5b917c47fec10547b65cf1d9096e194896fa1c2d6180536c8524289721a4c5b8
3
+ size 2333972488
CLAP/CLAP_weights_2023.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2cef4016d47d00eb28d153d522f397222057f95000e9bad6b9583c631284a1e6
3
+ size 689950036
CLAP/clapcap_weights_2023.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6d7b700f2bc1be7319fd60430a516a88f0277bf1158b4908687a14e6a2a3a006
3
+ size 1712037386
CLAP/config.yml ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # TEXT ENCODER CONFIG
2
+ text_model: 'bert-base-uncased'
3
+ text_len: 100
4
+ transformer_embed_dim: 768
5
+ freeze_text_encoder_weights: True
6
+
7
+ # AUDIO ENCODER CONFIG
8
+ audioenc_name: 'Cnn14'
9
+ out_emb: 2048
10
+ sampling_rate: 44100
11
+ duration: 9
12
+ fmin: 50
13
+ fmax: 14000
14
+ n_fft: 1028
15
+ hop_size: 320
16
+ mel_bins: 64
17
+ window_size: 1024
18
+
19
+ # PROJECTION SPACE CONFIG
20
+ d_proj: 1024
21
+ temperature: 0.003
22
+
23
+ # TRAINING AND EVALUATION CONFIG
24
+ num_classes: 527
25
+ batch_size: 1024
26
+ demo: False
bigvgan/args.yml ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ resblock: '1'
2
+ num_gpus: 0
3
+ batch_size: 64
4
+ num_mels: 80
5
+ learning_rate: 0.0001
6
+ adam_b1: 0.8
7
+ adam_b2: 0.99
8
+ lr_decay: 0.999
9
+ seed: 1234
10
+ upsample_rates:
11
+ - 4
12
+ - 4
13
+ - 2
14
+ - 2
15
+ - 2
16
+ - 2
17
+ upsample_kernel_sizes:
18
+ - 8
19
+ - 8
20
+ - 4
21
+ - 4
22
+ - 4
23
+ - 4
24
+ upsample_initial_channel: 1536
25
+ resblock_kernel_sizes:
26
+ - 3
27
+ - 7
28
+ - 11
29
+ resblock_dilation_sizes:
30
+ - - 1
31
+ - 3
32
+ - 5
33
+ - - 1
34
+ - 3
35
+ - 5
36
+ - - 1
37
+ - 3
38
+ - 5
39
+ activation: snakebeta
40
+ snake_logscale: true
41
+ resolutions:
42
+ - - 1024
43
+ - 120
44
+ - 600
45
+ - - 2048
46
+ - 240
47
+ - 1200
48
+ - - 512
49
+ - 50
50
+ - 240
51
+ mpd_reshapes:
52
+ - 2
53
+ - 3
54
+ - 5
55
+ - 7
56
+ - 11
57
+ use_spectral_norm: false
58
+ discriminator_channel_mult: 1
59
+ num_workers: 4
60
+ dist_config:
61
+ dist_backend: nccl
62
+ dist_url: tcp://localhost:54341
63
+ world_size: 1
bigvgan/best_netG.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:970ca75ee4d5ce583e9396a4534acb14971ea2b4f1c22e038f476680c868a789
3
+ size 449217313
maa1_caps.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1ce7630733efe17af5f9456fd5bed3f85337b4890773363a1028614f7e33510
3
+ size 2644760983
maa1_full.ckpt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:39b351b6499eab8fb5f7a3336a414633381dc963dc1621a62e8a188cc755713c
3
+ size 2644764506