egs/svc/VitsSVC/exp_config.json

{
    "base_config": "config/vitssvc.json",
    "model_type": "VitsSVC",
    "dataset": [
        "m4singer",
        "opencpop",
        "opensinger",
        "svcc",
        "vctk"
    ],
    "dataset_path": {
        // TODO: Fill in your dataset path
        "m4singer": "[M4Singer dataset path]",
        "opencpop": "[Opencpop dataset path]",
        "opensinger": "[OpenSinger dataset path]",
        "svcc": "[SVCC dataset path]",
        "vctk": "[VCTK dataset path]"
    },
    "use_custom_dataset": [],
    // TODO: Fill in the output log path. The default value is "Amphion/ckpts/svc"
    "log_dir": "ckpts/svc",
    "preprocess": {
        // TODO: Fill in the output data path. The default value is "Amphion/data"
        "processed_dir": "data",
        
        "n_mel": 100,
        "sample_rate": 24000,

        // contentvec
        "extract_contentvec_feature": true,
        "contentvec_sample_rate": 16000,
        "contentvec_batch_size": 1,
        "contentvec_frameshift": 0.02,
        // whisper
        "extract_whisper_feature": true,
        "whisper_sample_rate": 16000,
        "whisper_frameshift": 0.01,
        "whisper_downsample_rate": 2,
        // wenet
        "extract_wenet_feature": true,
        "wenet_downsample_rate": 4,
        "wenet_frameshift": 0.01,
        "wenet_sample_rate": 16000,
        // Fill in the content-based pretrained model's path
        "contentvec_file": "pretrained/contentvec/checkpoint_best_legacy_500.pt",
        "wenet_model_path": "pretrained/wenet/20220506_u2pp_conformer_exp/final.pt",
        "wenet_config": "pretrained/wenet/20220506_u2pp_conformer_exp/train.yaml",
        "whisper_model": "medium",
        "whisper_model_path": "pretrained/whisper/medium.pt",

        "use_contentvec": true,
        "use_whisper": true,
        "use_wenet": false,
        
        // Extract content features using dataloader
        "pin_memory": true,
        "num_workers": 8,
        "content_feature_batch_size": 16,

    },
    "model": {
        "condition_encoder": {
            // Config for features usage
            "merge_mode": "add",
            "use_log_loudness": true,
            "use_contentvec": true,
            "use_whisper": true,
            "use_wenet": false,
            "whisper_dim": 1024,
            "contentvec_dim": 256,
            "wenet_dim": 512,
        },
        "vits": {
            "inter_channels": 384,
            "hidden_channels": 384,
            "filter_channels": 256,
            "n_heads": 2,
            "n_layers": 6,
            "kernel_size": 3,
            "p_dropout": 0.1,
            "n_flow_layer": 4,
            "n_layers_q": 3,
            "gin_channels": 256,
            "n_speakers": 512,
            "use_spectral_norm": false,
        },
        "generator": "nsfhifigan",
    },
    "train": {
        "batch_size": 32,
        "learning_rate": 2e-4,
        "gradient_accumulation_step": 1,
        "max_epoch": -1, // -1 means no limit
        "save_checkpoint_stride": [
            3,
            50
        ],
        "keep_last": [
            3,
            2
        ],
    },
    "inference": {
        "batch_size": 1,
    }
}