32k.json

{
  "trainer": {
    "max_epochs": 20000,
    "accumulate_grad_batches": 1,
    "default_root_dir": "./logs_32k",
    "val_check_interval": 1.0
  },
  "train": {
    "log_interval": 200,
    "eval_interval": 1000,
    "seed": 1234,
    "max_epochs": 20000,
    "generator_learning_rate": 0.0001,
    "discriminator_learning_rate": 0.0001,
    "betas": [0.8, 0.99],
    "eps": 1e-9,
    "batch_size": 32,
    "fp16_run": true,
    "bp16_run": false,
    "lr_decay": 0.999875,
    "segment_size": 16384,
    "init_lr_ratio": 1,
    "warmup_epochs": 0,
    "c_mel": 45
  },
  "data": {
    "training_files": "filelists/32k_audio_filelist_train.txt",
    "validation_files": "filelists/32k_audio_filelist_valid.txt",
    "sampling_rate": 32000,
    "filter_length": 2048,
    "hop_length": 512,
    "win_length": 2048,
    "n_mel_channels": 128,
    "mel_fmin": 0.0,
    "mel_fmax": null
  },
  "model": {
    "inter_channels": 128,
    "resblock_kernel_sizes": [3,7,11,13],
    "resblock_dilation_sizes": [
      [1,3,5],
      [1,3,5],
      [1,3,5],
      [1,3,5]
    ],
    "upsample_rates": [8,8,4,2],
    "upsample_initial_channel": 512,
    "upsample_kernel_sizes": [16,16,8,4],
    "upsample_dilation_sizes": [1,1,1,1],
    "pre_kernel_size": 13,
    "post_kernel_size": 13,
    "use_spectral_norm": false,
    "multi_period_discriminator_periods": [2, 3, 5, 7, 11, 13, 17, 19, 23, 37]
  }
}