model.json

{
    "text_encoder": {
        "input_size": 512,
        "encoder_layers": 8,
        "encoder_heads": 2,
        "encoder_hidden": 384,
        "encoder_conv_kernel_size": [
            9,
            1
        ],
        "encoder_dropout": 0.2,
        "vocab_size": 366,
        "num_langs": 2,
        "num_speakers": 2,
        "max_position": 3072,
        "output_size": 512
    },
    "duration_predictor": {
        "input_size": 512,
        "hidden_size": 256,
        "k_size": 5,
        "layers": 3,
        "dropout_rate": 0.2
    },
    "mel_encoder": {
        "mel_bins": 100,
        "hidden_size": 512,
        "num_layers": 6,
        "kernel_size": 5,
        "dropout_rate": 0.2
    },
    "mel_decoder": {
        "mel_bins": 100,
        "hidden_size": 512,
        "num_layers": 6,
        "kernel_size": 5,
        "dropout_rate": 0.2
    },
    "post_net": {
        "n_mel_channels": 100,
        "postnet_embedding_dim": 512,
        "postnet_kernel_size": 9,
        "postnet_n_convolutions": 5
    },
    "wav_encoder": {
        "mel_bins": 100,
        "filter_length": 1024,
        "hop_length": 256,
        "win_length": 1024,
        "sampling_rate": 24000,
        "normalize": false,
        "power": 1,
        "norm": null,
        "center": true
    },
    "wav_decoder": {
        "vocos_model_id": "charactr/vocos-mel-24khz"
    },
    "delta": 0.2,
    "look_ahead": 3
}