train
dict | data
dict | model
dict | spk
dict |
|---|---|---|---|
{
"log_interval": 200,
"eval_interval": 800,
"seed": 1234,
"epochs": 10000,
"learning_rate": 0.0001,
"betas": [
0.8,
0.99
],
"eps": 1e-9,
"batch_size": 6,
"fp16_run": false,
"half_type": "fp16",
"lr_decay": 0.999875,
"segment_size": 10240,
"init_lr_ratio": 1,
"warmup_epochs": 0,
"c_mel": 45,
"c_kl": 1,
"use_sr": true,
"max_speclen": 512,
"port": "8001",
"keep_ckpts": 3,
"all_in_mem": false,
"vol_aug": true
} | {
"training_files": "filelists/train.txt",
"validation_files": "filelists/val.txt",
"max_wav_value": 32768,
"sampling_rate": 44100,
"filter_length": 2048,
"hop_length": 512,
"win_length": 2048,
"n_mel_channels": 80,
"mel_fmin": 0,
"mel_fmax": 22050,
"unit_interpolate_mode": "nearest"
} | {
"inter_channels": 192,
"hidden_channels": 192,
"filter_channels": 768,
"n_heads": 2,
"n_layers": 6,
"kernel_size": 3,
"p_dropout": 0.1,
"resblock": "1",
"resblock_kernel_sizes": [
3,
7,
11
],
"resblock_dilation_sizes": [
[
1,
3,
5
],
[
1,
3,
5
],
[
1,
3,
5
]
],
"upsample_rates": [
8,
8,
2,
2,
2
],
"upsample_initial_channel": 512,
"upsample_kernel_sizes": [
16,
16,
4,
4,
4
],
"n_layers_q": 3,
"n_flow_layer": 4,
"use_spectral_norm": false,
"gin_channels": 768,
"ssl_dim": 768,
"n_speakers": 1,
"vocoder_name": "nsf-hifigan",
"speech_encoder": "vec768l12",
"speaker_embedding": false,
"vol_embedding": true,
"use_depthwise_conv": false,
"flow_share_parameter": false,
"use_automatic_f0_prediction": true
} | {
"qiao": 0
} |