config.json

{
  "model": "xtts_v2",
  "output_path": "./output_model",
  "restore_path": null,
  "datasets": [
    {
      "formatter": "coqui",
      "meta_file_train": "./voxpopuli_es_500/metadata.json",
      "path": "./voxpopuli_es_500/wav_data",
      "language": "es"
    }
  ],
  "trainer": {
    "max_epochs": 5,
    "batch_size": 4,
    "save_step": 100,
    "log_step": 10,
    "eval_step": 100,
    "mixed_precision": true,
    "num_loader_workers": 4,
    "eval_split_max_size": 50,
    "run_eval": true
  },
  "audio": {
    "sample_rate": 22050,
    "win_length": 1024,
    "hop_length": 256,
    "fft_size": 1024,
    "mel_fmin": 0,
    "mel_fmax": 8000
  },
  "optimizer": {
    "type": "adam",
    "lr": 0.0001,
    "weight_decay": 0.0
  },
  "lr_scheduler": {
    "type": "step",
    "step_size": 1000,
    "gamma": 0.5
  },
  "use_cuda": true,
  "seed": 42
}