config.json

{
  "sample_rate": 24000,
  "backbone_class_path": "exp.models.ComVo",
  "backbone_init_args": {
    "input_channels": 100,
    "dim": 512,
    "intermediate_dim": 1536,
    "num_layers": 8,
    "n_quantization": 128,
    "layer_scale_init_value": null,
    "adanorm_num_embeddings": null,
    "rank": null
  },
  "head_class_path": "exp.heads.ISTFTHead",
  "head_init_args": {
    "dim": 512,
    "n_fft": 1024,
    "hop_length": 256,
    "padding": "center"
  },
  "feature_extractor_class_path": "exp.feature_extractors.MelSpectrogramFeatures",
  "feature_extractor_init_args": {
    "sample_rate": 24000,
    "n_fft": 1024,
    "hop_length": 256,
    "n_mels": 100,
    "padding": "center"
  }
}