vqvae-speech/architecture.json at master · JeremyCCHsu/vqvae-speech · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
{
    "T": 8192,
    "fs": 16000,
    "num_symbol": 256,
    "dim_symbol_emb": 1,
    "num_exemplar": 512,
    "dim_exemplar": 64,
    "num_speaker": 109,
    "dim_speaker": 64,
    "n_emb_project": 4096,
    "n_downsample_stack": 6,
    "initial_filter_width": 32,
    "filter_width": 2,
    "dilations": [1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
                  1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
                  1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
                  1, 2, 4, 8, 16, 32, 64, 128, 256, 512,
                  1, 2, 4, 8, 16, 32, 64, 128, 256, 512],
    "residual_channels": 128,
    "dilation_channels": 128,
    "skip_channels": 1028,
    "training": {
        "lr": 1e-4,
        "beta1": 0.5,
        "beta2": 0.999,
        "batch_size": 4,
        "refresh_freq": 500,
        "nIterD": 5,
        "lambda": 10,
        "co_coeff": 10.0,
        "nIterPretrain": 30000,
        "intenseIter": 50000,
        "maxIter": 1000000
    },
    "discriminator": {
      "output": [128, 128],
      "lipschitz": 1.0
    },
    "generator": {
      "input": 64,
      "output": [128, 128]
    },
    "encoder": {
        "kernels": 5,
        "filters": 128
    }
}