{
  "_class_name": "LTX2VocoderWithBWE",
  "_diffusers_version": "0.37.0.dev0",
  "act_fn": "snakebeta",
  "antialias": true,
  "antialias_kernel_size": 12,
  "antialias_ratio": 2,
  "bwe_act_fn": "snakebeta",
  "bwe_antialias": true,
  "bwe_antialias_kernel_size": 12,
  "bwe_antialias_ratio": 2,
  "bwe_final_act_fn": null,
  "bwe_final_bias": false,
  "bwe_hidden_channels": 512,
  "bwe_in_channels": 128,
  "bwe_leaky_relu_negative_slope": 0.1,
  "bwe_out_channels": 2,
  "bwe_resnet_dilations": [
    [
      1,
      3,
      5
    ],
    [
      1,
      3,
      5
    ],
    [
      1,
      3,
      5
    ]
  ],
  "bwe_resnet_kernel_sizes": [
    3,
    7,
    11
  ],
  "bwe_upsample_factors": [
    6,
    5,
    2,
    2,
    2
  ],
  "bwe_upsample_kernel_sizes": [
    12,
    11,
    4,
    4,
    4
  ],
  "filter_length": 512,
  "final_act_fn": null,
  "final_bias": false,
  "hidden_channels": 1536,
  "hop_length": 80,
  "in_channels": 128,
  "input_sampling_rate": 16000,
  "leaky_relu_negative_slope": 0.1,
  "num_mel_channels": 64,
  "out_channels": 2,
  "output_sampling_rate": 48000,
  "resnet_dilations": [
    [
      1,
      3,
      5
    ],
    [
      1,
      3,
      5
    ],
    [
      1,
      3,
      5
    ]
  ],
  "resnet_kernel_sizes": [
    3,
    7,
    11
  ],
  "upsample_factors": [
    5,
    2,
    2,
    2,
    2,
    2
  ],
  "upsample_kernel_sizes": [
    11,
    4,
    4,
    4,
    4,
    4
  ],
  "window_length": 512
}