RMSnow's picture
init and interface
df2accb
raw
history blame
6.08 kB
{
"base_config": "egs/vocoder/gan/exp_config_base.json",
"exp_name": "bigvgan_large",
"inference": {
"batch_size": 1,
},
"model": {
"bigvgan": {
"activation": "snakebeta",
"resblock": "1",
"resblock_dilation_sizes": [
[
1,
3,
5,
],
[
1,
3,
5,
],
[
1,
3,
5,
],
],
"resblock_kernel_sizes": [
3,
7,
11,
],
"snake_logscale": true,
"upsample_initial_channel": 1536,
"upsample_kernel_sizes": [
8,
8,
4,
4,
4,
4,
],
"upsample_rates": [
4,
4,
2,
2,
2,
2,
],
},
"discriminators": [
"mpd",
"msstftd",
],
"generator": "bigvgan",
"mpd": {
"discriminator_channel_multi": 1,
"mpd_reshapes": [
2,
3,
5,
7,
11,
],
"use_spectral_norm": false,
},
"mrd": {
"discriminator_channel_multi": 1,
"mrd_override": false,
"resolutions": [
[
1024,
120,
600,
],
[
2048,
240,
1200,
],
[
512,
50,
240,
],
],
"use_spectral_norm": false,
},
"msstftd": {
"filters": 32,
},
},
"model_type": "GANVocoder",
"preprocess": {
"audio_dir": "audios",
"bits": 8,
"contentvec_dir": "contentvec",
"cut_mel_frame": 32,
"data_augment": false,
"dur_dir": "durs",
"duration_dir": "duration",
"emo2id": "emo2id.json",
"energy_dir": "energys",
"energy_extract_mode": "from_mel",
"energy_norm": false,
"extract_audio": true,
"extract_contentvec_feature": false,
"extract_duration": false,
"extract_energy": false,
"extract_label": false,
"extract_mcep": false,
"extract_mel": true,
"extract_mert_feature": false,
"extract_one_hot": false,
"extract_pitch": false,
"extract_uv": false,
"extract_wenet_feature": false,
"extract_whisper_feature": false,
"f0_max": 1100,
"f0_min": 50,
"file_lst": "file.lst",
"fmax": 12000,
"fmin": 0,
"hop_size": 256,
"is_mu_law": false,
"lab_dir": "labs",
"label_dir": "labels",
"mcep_dir": "mcep",
"mel_dir": "mels",
"mel_min_max_norm": false,
"min_level_db": -115,
"n_fft": 1024,
"n_mel": 100,
"num_silent_frames": 8,
"phone_seq_file": "phone_seq_file",
"pitch_bin": 256,
"pitch_dir": "pitches",
"pitch_extractor": "parselmouth",
"pitch_max": 1100.0,
"pitch_min": 50.0,
"pitch_norm": false,
"processed_dir": "processed_data",
"ref_level_db": 20,
"sample_rate": 24000,
"spk2id": "singers.json",
"train_file": "train.json",
"trim_fft_size": 512,
"trim_hop_size": 128,
"trim_silence": false,
"trim_top_db": 30,
"trimmed_wav_dir": "trimmed_wavs",
"use_audio": true,
"use_dur": false,
"use_emoid": false,
"use_frame_duration": false,
"use_frame_energy": false,
"use_frame_pitch": false,
"use_lab": false,
"use_label": false,
"use_log_scale_energy": false,
"use_log_scale_pitch": false,
"use_mel": true,
"use_one_hot": false,
"use_phn_seq": false,
"use_phone_duration": false,
"use_phone_energy": false,
"use_phone_pitch": false,
"use_spkid": false,
"use_uv": false,
"use_wav": false,
"use_wenet": false,
"utt2emo": "utt2emo",
"utt2spk": "utt2spk",
"uv_dir": "uvs",
"valid_file": "test.json",
"wav_dir": "wavs",
"wenet_dir": "wenet",
"win_size": 1024,
},
"supported_model_type": [
"GANVocoder",
"Fastspeech2",
"DiffSVC",
"Transformer",
"EDM",
"CD",
],
"train": {
"adamw": {
"adam_b1": 0.8,
"adam_b2": 0.99,
"lr": 0.0002,
},
"batch_size": 4,
"criterions": [
"feature",
"discriminator",
"generator",
"mel",
],
"dataloader": {
"num_worker": 4,
"pin_memory": true,
},
"ddp": true,
"epochs": 50000,
"exponential_lr": {
"lr_decay": 0.999,
},
"gradient_accumulation_step": 1,
"keep_checkpoint_max": 5,
"max_epoch": 1000000,
"max_steps": 1000000,
"multi_speaker_training": false,
"random_seed": 114514,
"run_eval": [
true,
],
"sampler": {
"drop_last": true,
"holistic_shuffle": true,
},
"save_checkpoint_stride": [
200,
],
"save_checkpoints_steps": 10000,
"save_summary_steps": 500,
"total_training_steps": 50000,
"tracker": [
"tensorboard",
],
"valid_interval": 10000,
},
}