config.yaml

Model:
  SpeakerEncoder:
    c_in: 80
    c_h: 128
    c_out: 128
    kernel_size: 5
    bank_size: 8
    bank_scale: 1
    c_bank: 128
    n_conv_blocks: 6
    n_dense_blocks: 6
    subsample: [1, 2, 1, 2, 1, 2]
    act: "relu"
    dropout_rate: 0.0
  ContentEncoder:
    c_in: 80
    c_h: 128
    c_out: 128
    kernel_size: 5
    bank_size: 8
    bank_scale: 1
    c_bank: 128
    n_conv_blocks: 6
    subsample: [1, 2, 1, 2, 1, 2]
    act: "relu"
    dropout_rate: 0.0
  Decoder:
    c_in: 128
    c_cond: 128
    c_h: 128
    c_out: 80
    kernel_size: 5
    n_conv_blocks: 6
    upsample: [2, 1, 2, 1, 2, 1]
    act: "relu"
    sn: False
    dropout_rate: 0.0
Optimizer:
  lr: 0.0005
  beta1: 0.9
  beta2: 0.999
  amsgrad: True
  weight_decay: 0.0001
  grad_norm: 5
Lambda:
  rec: 10
  kl: 1
  kl_annealing: 20000