{
  "model_type": "autoencoder",
  "sample_size": 96000,
  "sample_rate": 48000,
  "audio_channels": 1,
  "model": {
    "encoder": {
      "type": "spec_convnext",
      "config": {
        "in_channels": 1,
        "channels": 512,
        "c_mults": [1, 2, 4],
        "strides": [4, 5, 6],
        "latent_dim": 1024,
        "identity_init": false,
        "n_fft": 64,
        "hop_length": 16,
        "use_snake": true,
        "causal": true,
        "padding_mode": "reflect"
      }
    },
    "decoder": {
      "type": "oobleck",
      "config": {
        "out_channels": 1,
        "channels": 320,
        "c_mults": [1, 2, 4, 8, 16],
        "strides": [2, 4, 5, 6, 8],
        "latent_dim": 512,
        "use_snake": true,
        "final_tanh": false,
        "causal": true,
        "padding_mode": "reflect"
      }
    },
    "bottleneck": {
      "type": "vae_distillation",
      "config": {
        "latent_dim": 512,
        "distillation_proj_dim": 1024
      }
    },
    "latent_dim": 512,
    "downsampling_ratio": 1920,
    "io_channels": 1
  }
}