| config: egs/LibriTTS/tokenizer/conf/encodec_lstm_16k_n32_600k_step_rmseg_use_power_ds640.yaml | |
| print_config: false | |
| log_level: INFO | |
| dry_run: false | |
| iterator_type: sequence | |
| output_dir: egs/LibriTTS/tokenizer/exp/encodec_lstm_16k_n32_600k_step_rmseg_use_power_ds640_raw_en_libritts | |
| ngpu: 1 | |
| seed: 0 | |
| num_workers: 8 | |
| num_att_plot: 0 | |
| dist_backend: nccl | |
| dist_init_method: env:// | |
| dist_world_size: null | |
| dist_rank: null | |
| local_rank: 0 | |
| dist_master_addr: null | |
| dist_master_port: null | |
| dist_launcher: null | |
| multiprocessing_distributed: true | |
| unused_parameters: true | |
| sharded_ddp: false | |
| cudnn_enabled: true | |
| cudnn_benchmark: false | |
| cudnn_deterministic: false | |
| collect_stats: false | |
| write_collected_feats: false | |
| max_epoch: 60 | |
| max_update: 9223372036854775807 | |
| patience: null | |
| val_scheduler_criterion: | |
| - valid | |
| - loss | |
| early_stopping_criterion: | |
| - valid | |
| - loss | |
| - min | |
| best_model_criterion: | |
| - - valid | |
| - generator_multi_spectral_recon_loss | |
| - min | |
| keep_nbest_models: 60 | |
| nbest_averaging_interval: 0 | |
| grad_clip: -1 | |
| grad_clip_type: 2.0 | |
| grad_noise: false | |
| accum_grad: 1 | |
| no_forward_run: false | |
| resume: true | |
| train_dtype: float32 | |
| use_amp: false | |
| log_interval: 50 | |
| use_tensorboard: true | |
| use_wandb: false | |
| wandb_project: null | |
| wandb_id: null | |
| wandb_entity: null | |
| wandb_name: null | |
| wandb_model_log_interval: -1 | |
| detect_anomaly: false | |
| pretrain_path: null | |
| init_param: [] | |
| ignore_init_mismatch: true | |
| freeze_param: [] | |
| num_iters_per_epoch: 10000 | |
| batch_size: 32 | |
| valid_batch_size: null | |
| batch_bins: 1000000 | |
| valid_batch_bins: null | |
| drop_last: true | |
| train_shape_file: | |
| - exp/tokenizer_states_16k/train/speech_shape | |
| valid_shape_file: | |
| - exp/tokenizer_states_16k/dev/speech_shape | |
| batch_type: unsorted | |
| valid_batch_type: null | |
| speech_length_min: -1 | |
| speech_length_max: -1 | |
| fold_length: | |
| - 512 | |
| - 150 | |
| sort_in_batch: descending | |
| sort_batch: descending | |
| multiple_iterator: false | |
| chunk_length: 500 | |
| chunk_shift_ratio: 0.5 | |
| num_cache_chunks: 1024 | |
| dataset_type: small | |
| dataset_conf: {} | |
| train_data_file: null | |
| valid_data_file: null | |
| train_data_path_and_name_and_type: | |
| - - dump/raw_16k/train/wav.scp.pai | |
| - speech | |
| - kaldi_ark | |
| valid_data_path_and_name_and_type: | |
| - - dump/raw_16k/dev/wav.scp.pai | |
| - speech | |
| - kaldi_ark | |
| allow_variable_data_keys: false | |
| max_cache_size: 0.0 | |
| max_cache_fd: 32 | |
| valid_max_cache_size: null | |
| optim: adam | |
| optim_conf: | |
| lr: 0.0003 | |
| betas: | |
| - 0.5 | |
| - 0.9 | |
| scheduler: null | |
| scheduler_conf: | |
| step_size: 8 | |
| gamma: 0.1 | |
| optim2: adam | |
| optim2_conf: | |
| lr: 0.0003 | |
| betas: | |
| - 0.5 | |
| - 0.9 | |
| scheduler2: null | |
| scheduler2_conf: | |
| step_size: 8 | |
| gamma: 0.1 | |
| simple_ddp: false | |
| num_worker_count: 1 | |
| generator_first: false | |
| input_size: 1 | |
| cmvn_file: null | |
| disc_grad_clip: -1 | |
| disc_grad_clip_type: 2.0 | |
| gen_train_interval: 1 | |
| disc_train_interval: 1 | |
| use_preprocessor: true | |
| speech_volume_normalize: null | |
| speech_rms_normalize: false | |
| speech_max_length: 51200 | |
| sampling_rate: 16000 | |
| valid_max_length: 51200 | |
| frontend: null | |
| frontend_conf: {} | |
| normalize: null | |
| normalize_conf: {} | |
| encoder: encodec_seanet_encoder | |
| encoder_conf: | |
| ratios: | |
| - 8 | |
| - 5 | |
| - 4 | |
| - 2 | |
| - 2 | |
| norm: time_group_norm | |
| causal: false | |
| quantizer: costume_quantizer | |
| quantizer_conf: | |
| codebook_size: 1024 | |
| num_quantizers: 32 | |
| ema_decay: 0.99 | |
| kmeans_init: true | |
| sampling_rate: 16000 | |
| quantize_dropout: true | |
| rand_num_quant: | |
| - 1 | |
| - 2 | |
| - 4 | |
| - 8 | |
| - 16 | |
| - 32 | |
| use_ddp: true | |
| encoder_hop_length: 640 | |
| decoder: encodec_seanet_decoder | |
| decoder_conf: | |
| ratios: | |
| - 8 | |
| - 5 | |
| - 4 | |
| - 2 | |
| - 2 | |
| norm: time_group_norm | |
| causal: false | |
| model: encodec | |
| model_conf: | |
| odim: 128 | |
| multi_spectral_window_powers_of_two: | |
| - 5 | |
| - 6 | |
| - 7 | |
| - 8 | |
| - 9 | |
| - 10 | |
| target_sample_hz: 16000 | |
| audio_normalize: true | |
| segment_dur: null | |
| overlap_ratio: null | |
| use_power_spec_loss: true | |
| discriminator: multiple_disc | |
| discriminator_conf: | |
| disc_conf_list: | |
| - filters: 32 | |
| name: encodec_multi_scale_stft_discriminator | |
| distributed: false | |
| version: 0.2.0 | |