| data: | |
| f0_extractor: 'parselmouth' # 'parselmouth', 'dio', 'harvest', or 'crepe' | |
| f0_min: 65 # about C2 | |
| f0_max: 1600 # about G5 | |
| sampling_rate: 44100 | |
| block_size: 512 # Equal to hop_length | |
| duration: 2 # Audio duration during training, must be less than the duration of the shortest audio clip | |
| encoder: 'hubertsoft' # 'hubertsoft', 'hubertbase' or 'contentvec' | |
| encoder_sample_rate: 16000 | |
| encoder_hop_size: 320 | |
| encoder_out_channels: 256 | |
| encoder_ckpt: pretrain/hubert/hubert-soft-0d54a1f4.pt | |
| train_path: data/train # Create a folder named "audio" under this path and put the audio clip in it | |
| valid_path: data/val # Create a folder named "audio" under this path and put the audio clip in it | |
| model: | |
| type: 'CombSubFast' | |
| n_spk: 1 # max number of different speakers | |
| enhancer: | |
| type: 'nsf-hifigan' | |
| ckpt: 'pretrain/nsf_hifigan/model' | |
| loss: | |
| fft_min: 256 | |
| fft_max: 2048 | |
| n_scale: 4 # rss kernel numbers | |
| device: cuda | |
| env: | |
| expdir: exp/combsub-test | |
| gpu_id: 0 | |
| train: | |
| num_workers: 2 # If your cpu and gpu are both very strong, set to 0 may be faster! | |
| batch_size: 24 | |
| cache_all_data: true # Save Internal-Memory or Graphics-Memory if it is false, but may be slow | |
| cache_device: 'cpu' # Set to 'cuda' to cache the data into the Graphics-Memory, fastest speed for strong gpu | |
| cache_fp16: true | |
| epochs: 100000 | |
| interval_log: 10 | |
| interval_val: 2000 | |
| lr: 0.0005 | |
| weight_decay: 0 |