mini_config.yaml

log_dir: "Models/Output"
save_freq: 5
log_interval: 10
device: "cuda"
epochs: 50
batch_size: 8
max_len: 400
pretrained_model: ""
second_stage_load_pretrained: true
load_only_params: true

external_models:
  asr:
    input_dim: 80
    hidden_dim: 256
    n_token: 178
  plbert:
    vocab_size: 178
    hidden_size: 768
    num_attention_heads: 12
    intermediate_size: 2048
    dropout: 0.1

data_params:
  train_data: "shethjenil/audiodata"
  root_path: ""
  min_length: 50

preprocess_params:
  sr: 24000
  n_fft: 2048
  win_length: 1200
  hop_length: 300

model_params:
  multispeaker: true
  dim_in: 64
  hidden_dim: 128
  max_conv_dim: 512
  n_layer: 2
  n_mels: 80
  n_token: 178
  max_dur: 50
  style_dim: 128
  dropout: 0.2
  decoder:
    type: "istftnet"
    hidden_dim: 256
    decoder_out_dim: 256
    asr_res_in: 128
    resblock_kernel_sizes: [3, 3]
    upsample_rates: [10, 6]
    upsample_initial_channel: 256
    resblock_dilation_sizes: [[1, 3, 5], [1, 3, 5], [1, 3, 5], [1, 3, 5]]
    upsample_kernel_sizes: [20, 12]
    gen_istft_n_fft: 20
    gen_istft_hop_size: 5
    disable_complex: true
  slm:
    model: "microsoft/wavlm-base-plus"
    sr: 16000
    hidden: 768
    nlayers: 13
    initial_channel: 64
  diffusion:
    embedding_mask_proba: 0.1
    transformer:
      num_layers: 3
      num_heads: 8
      head_features: 64
      multiplier: 2
    dist:
      sigma_data: 0.2
      estimate_sigma_data: true
      mean: -3.0
      std: 1.0

loss_params:
  lambda_mel: 5.0
  lambda_gen: 1.0
  lambda_slm: 1.0
  lambda_mono: 1.0
  lambda_s2s: 1.0
  lambda_f0: 1.0
  lambda_norm: 1.0
  lambda_dur: 1.0
  lambda_ce: 20.0
  lambda_sty: 1.0
  lambda_diff: 1.0
  diff_epoch: 10
  joint_epoch: 30

optimizer_params:
  lr: 0.0001
  bert_lr: 0.00001
  ft_lr: 0.0001

slmadv_params:
  min_len: 400
  max_len: 500
  batch_percentage: 0.5
  iter: 10
  thresh: 5.0
  scale: 0.01
  sig: 1.5