| acous_params: | |
| - - 480 | |
| - 1200 | |
| - 80 | |
| - - 240 | |
| - 1200 | |
| - 160 | |
| amp: false | |
| audio_num_mel_bins: 160 | |
| audio_sample_rate: 24000 | |
| c_spk_enc: 512 | |
| char_dict_size: 15000 | |
| conv_use_pos: false | |
| dec0_dilations: | |
| - 1 | |
| - 2 | |
| - 4 | |
| - 1 | |
| - 2 | |
| - 4 | |
| - 1 | |
| dec0_kernel_size: 3 | |
| dec_dilations: | |
| - 1 | |
| - 2 | |
| - 1 | |
| - 2 | |
| - 1 | |
| dec_ffn_kernel_size: 9 | |
| dec_inp_add_noise: false | |
| dec_kernel_size: 5 | |
| dec_layers: 4 | |
| dec_post_net_kernel: 3 | |
| decoder_rnn_dim: 0 | |
| decoder_type: conv | |
| dropout: 0.0 | |
| ds_add_pitch_embed: false | |
| dur_alpha: 1.0 | |
| dur_context_enc: true | |
| dur_log: true | |
| dur_predictor_kernel: 3 | |
| dur_predictor_layers: 2 | |
| dur_use_char: true | |
| dur_use_spk: true | |
| enc_dec_norm: ln | |
| enc_dilations: | |
| - 1 | |
| - 1 | |
| - 1 | |
| - 1 | |
| enc_ffn_kernel_size: 5 | |
| enc_kernel_size: 5 | |
| enc_layers: 8 | |
| enc_post_net_kernel: 3 | |
| enc_pre_ln: true | |
| enc_prenet: true | |
| encoder_K: 8 | |
| encoder_type: rel_fft | |
| f0_max: 600 | |
| f0_min: 60 | |
| ffn_act: gelu | |
| ffn_hidden_size: 1024 | |
| fft_size: 1200 | |
| fg_spk_enc_hidden: 256 | |
| fmax: 12000 | |
| fmin: 0 | |
| frames_multiple: 8 | |
| hidden_size: 512 | |
| hop_size: 240 | |
| ignore_begin_end_sil: false | |
| keep_c0_init: true | |
| kl_min: 0 | |
| kl_start_steps: 1 | |
| lat_for_dur: false | |
| latent_dim: 16 | |
| latent_size: 256 | |
| layers_in_block: 2 | |
| ling_label_dict_size: | |
| - 20 | |
| - 4 | |
| - 5 | |
| - 2 | |
| - 3 | |
| - 3 | |
| - 3 | |
| - 6 | |
| - 15 | |
| ling_labels: | |
| - tone | |
| load_ckpt: '' | |
| loud_norm: false | |
| mel_vmax: 0.5 | |
| mel_vmin: -6 | |
| min_frames: 50 | |
| mixed_precision: bf16 | |
| no_text_enc: false | |
| nsf_type: none | |
| num_heads: 2 | |
| out_wav_norm: true | |
| pad_frames: false | |
| precision: fp16 | |
| predict_pitch: false | |
| resblock: '1' | |
| resblock_dilation_sizes: | |
| - - 1 | |
| - 3 | |
| - 5 | |
| - - 1 | |
| - 3 | |
| - 5 | |
| - - 1 | |
| - 3 | |
| - 5 | |
| resblock_kernel_sizes: | |
| - 3 | |
| - 7 | |
| - 11 | |
| train_spk_embed_only: false | |
| upsample_initial_channel: 512 | |
| upsample_kernel_sizes: | |
| - 12 | |
| - 11 | |
| - 8 | |
| - 4 | |
| upsample_rates: | |
| - 6 | |
| - 5 | |
| - 4 | |
| - 2 | |
| use_bert_input: false | |
| use_cfg: true | |
| use_char: true | |
| use_cur_global: false | |
| use_cur_global_dec: true | |
| use_dur_embed: true | |
| use_dur_mask_embed: true | |
| use_ema: false | |
| use_expand_ph: true | |
| use_finegrained_spk: false | |
| use_global_lat: false | |
| use_gt_dur: false | |
| use_gt_f0: false | |
| use_mix_spk_embed: false | |
| use_new_vae: false | |
| use_ph_level_f0: false | |
| use_ph_pos_embed: true | |
| use_pitch_embed: false | |
| use_pitch_embed_dec: false | |
| use_pitch_pred: true | |
| use_pos_embed: true | |
| use_qk_norm: true | |
| use_random_spk_embed: false | |
| use_seq_cfg: true | |
| use_spk_embed: false | |
| use_spk_enc: true | |
| use_spk_id: false | |
| use_uv: true | |
| use_vae: true | |
| use_vpcfm: true | |
| use_vqvae: true | |
| use_word_encoder: true | |
| use_word_input: false | |
| vae_dur_grad: 0.1 | |
| vae_enc_hidden_size: 384 | |
| vae_stride: 4 | |
| vae_word_conder_layers: 0 | |
| vq_stride: 8 | |
| win_size: 1200 | |
| word_dict_size: 10000 | |
| melgan_config: | |
| all_noise: false | |
| backbone_resampling: librosa_kaiser_best | |
| batch_size: 8 | |
| cond_disc: false | |
| dim_pitch_condition: 1 | |
| downsamp_factor: 4 | |
| epochs: 1000 | |
| frame_shift: 240 | |
| lambda_feat: 0.0 | |
| lambda_log_pitch: 0.4 | |
| lambda_voiced: 1.0 | |
| load_D: 1 | |
| log_interval: 100 | |
| loss_pitch: 1.0 | |
| loss_speaker: 1.0 | |
| loss_stft: 0.0 | |
| lr: 0.0005 | |
| mode_pitch_condition: singgan_torch | |
| multi_resolution: 0 | |
| n_layers_D: 4 | |
| n_mel_channels: 160 | |
| n_residual_layers: 4 | |
| n_test_samples: 5 | |
| ndf: 16 | |
| noise_index: 1.0 | |
| nr: 0 | |
| num_D: 3 | |
| num_band: 1 | |
| num_workers: 0 | |
| offset: 0 | |
| pretrain_steps: 0 | |
| res_layers: 1 | |
| run_hdfs: 0 | |
| sampling_rate: 24000 | |
| save_interval: 5000 | |
| seq_len: 100 | |
| single_stft: 0 | |
| sub_dis: 1 | |
| tf: 1 | |
| tf_end_ratio: 0.0 | |
| tf_end_step: 0 | |
| tf_start_ratio: 0.0 | |
| tf_start_step: 0 | |
| up_sample: | |
| - 5 | |
| - 4 | |
| - 4 | |
| - 3 | |
| use_F_dis: 0 | |
| use_aug_pitch: 0 | |
| use_interpolate: 0 | |
| use_lsgan: 1 | |
| use_mel_loss: 1 | |
| use_melnorm: 0 | |
| use_msg_gan: 0 | |
| use_pitch_condition: false | |
| use_pitch_prediction: 1 | |
| use_sbd: 0 | |
| use_speaker_prediction: 0 | |
| use_tanh: true | |
| use_time_loss: 1 |