| { |
| "log_dir": "./runs/run_dit_mel_seed_uvit_whisper_base_f0_44k", |
| "save_freq": 1, |
| "log_interval": 10, |
| "save_interval": 1000, |
| "device": "cuda", |
| "epochs": 1000, |
| "batch_size": 1, |
| "batch_length": 100, |
| "max_len": 80, |
| "pretrained_model": "", |
| "pretrained_encoder": "", |
| "load_only_params": false, |
| "preprocess_params": { |
| "sr": 44100, |
| "spect_params": { |
| "n_fft": 2048, |
| "win_length": 2048, |
| "hop_length": 512, |
| "n_mels": 128, |
| "fmin": 0, |
| "fmax": "None" |
| } |
| }, |
| "model_params": { |
| "dit_type": "DiT", |
| "reg_loss_type": "l1", |
| "timbre_shifter": { |
| "se_db_path": "./modules/openvoice/checkpoints_v2/converter/se_db.pt", |
| "ckpt_path": "./modules/openvoice/checkpoints_v2/converter" |
| }, |
| "vocoder": { |
| "type": "bigvgan", |
| "name": "nvidia/bigvgan_v2_44khz_128band_512x" |
| }, |
| "speech_tokenizer": { |
| "type": "whisper", |
| "name": "openai/whisper-small" |
| }, |
| "style_encoder": { |
| "dim": 192, |
| "campplus_path": "campplus_cn_common.bin" |
| }, |
| "DAC": { |
| "encoder_dim": 64, |
| "encoder_rates": [ |
| 2, |
| 5, |
| 5, |
| 6 |
| ], |
| "decoder_dim": 1536, |
| "decoder_rates": [ |
| 6, |
| 5, |
| 5, |
| 2 |
| ], |
| "sr": 24000 |
| }, |
| "length_regulator": { |
| "channels": 768, |
| "is_discrete": false, |
| "in_channels": 768, |
| "content_codebook_size": 2048, |
| "sampling_ratios": [ |
| 1, |
| 1, |
| 1, |
| 1 |
| ], |
| "vector_quantize": false, |
| "n_codebooks": 1, |
| "quantizer_dropout": 0.0, |
| "f0_condition": true, |
| "n_f0_bins": 256 |
| }, |
| "DiT": { |
| "hidden_dim": 768, |
| "num_heads": 12, |
| "depth": 17, |
| "class_dropout_prob": 0.1, |
| "block_size": 8192, |
| "in_channels": 128, |
| "style_condition": true, |
| "final_layer_type": "mlp", |
| "target": "mel", |
| "content_dim": 768, |
| "content_codebook_size": 1024, |
| "content_type": "discrete", |
| "f0_condition": true, |
| "n_f0_bins": 256, |
| "content_codebooks": 1, |
| "is_causal": false, |
| "long_skip_connection": false, |
| "zero_prompt_speech_token": false, |
| "time_as_token": false, |
| "style_as_token": false, |
| "uvit_skip_connection": true, |
| "add_resblock_in_transformer": false |
| }, |
| "wavenet": { |
| "hidden_dim": 768, |
| "num_layers": 8, |
| "kernel_size": 5, |
| "dilation_rate": 1, |
| "p_dropout": 0.2, |
| "style_condition": true |
| } |
| }, |
| "loss_params": { |
| "base_lr": 0.0001, |
| "lambda_mel": 45, |
| "lambda_kl": 1.0 |
| } |
| } |
|
|
