config.json
| 1 | { |
| 2 | "output_path": "output", |
| 3 | "logger_uri": null, |
| 4 | "run_name": "run", |
| 5 | "project_name": null, |
| 6 | "run_description": "\ud83d\udc38Coqui trainer run.", |
| 7 | "print_step": 25, |
| 8 | "plot_step": 100, |
| 9 | "model_param_stats": false, |
| 10 | "wandb_entity": null, |
| 11 | "dashboard_logger": "tensorboard", |
| 12 | "save_on_interrupt": true, |
| 13 | "log_model_step": null, |
| 14 | "save_step": 10000, |
| 15 | "save_n_checkpoints": 5, |
| 16 | "save_checkpoints": true, |
| 17 | "save_all_best": false, |
| 18 | "save_best_after": 10000, |
| 19 | "target_loss": null, |
| 20 | "print_eval": false, |
| 21 | "test_delay_epochs": 0, |
| 22 | "run_eval": true, |
| 23 | "run_eval_steps": null, |
| 24 | "distributed_backend": "nccl", |
| 25 | "distributed_url": "tcp://localhost:54321", |
| 26 | "mixed_precision": false, |
| 27 | "precision": "fp16", |
| 28 | "epochs": 1000, |
| 29 | "batch_size": 32, |
| 30 | "eval_batch_size": 16, |
| 31 | "grad_clip": 0.0, |
| 32 | "scheduler_after_epoch": true, |
| 33 | "lr": 0.001, |
| 34 | "optimizer": "radam", |
| 35 | "optimizer_params": null, |
| 36 | "lr_scheduler": null, |
| 37 | "lr_scheduler_params": {}, |
| 38 | "use_grad_scaler": false, |
| 39 | "allow_tf32": false, |
| 40 | "cudnn_enable": true, |
| 41 | "cudnn_deterministic": false, |
| 42 | "cudnn_benchmark": false, |
| 43 | "training_seed": 54321, |
| 44 | "model": "xtts", |
| 45 | "num_loader_workers": 0, |
| 46 | "num_eval_loader_workers": 0, |
| 47 | "use_noise_augment": false, |
| 48 | "audio": { |
| 49 | "sample_rate": 22050, |
| 50 | "output_sample_rate": 24000 |
| 51 | }, |
| 52 | "use_phonemes": false, |
| 53 | "phonemizer": null, |
| 54 | "phoneme_language": null, |
| 55 | "compute_input_seq_cache": false, |
| 56 | "text_cleaner": null, |
| 57 | "enable_eos_bos_chars": false, |
| 58 | "test_sentences_file": "", |
| 59 | "phoneme_cache_path": null, |
| 60 | "characters": null, |
| 61 | "add_blank": false, |
| 62 | "batch_group_size": 0, |
| 63 | "loss_masking": null, |
| 64 | "min_audio_len": 1, |
| 65 | "max_audio_len": Infinity, |
| 66 | "min_text_len": 1, |
| 67 | "max_text_len": Infinity, |
| 68 | "compute_f0": false, |
| 69 | "compute_energy": false, |
| 70 | "compute_linear_spec": false, |
| 71 | "precompute_num_workers": 0, |
| 72 | "start_by_longest": false, |
| 73 | "shuffle": false, |
| 74 | "drop_last": false, |
| 75 | "datasets": [ |
| 76 | { |
| 77 | "formatter": "", |
| 78 | "dataset_name": "", |
| 79 | "path": "", |
| 80 | "meta_file_train": "", |
| 81 | "ignored_speakers": null, |
| 82 | "language": "", |
| 83 | "phonemizer": "", |
| 84 | "meta_file_val": "", |
| 85 | "meta_file_attn_mask": "" |
| 86 | } |
| 87 | ], |
| 88 | "test_sentences": [], |
| 89 | "eval_split_max_size": null, |
| 90 | "eval_split_size": 0.01, |
| 91 | "use_speaker_weighted_sampler": false, |
| 92 | "speaker_weighted_sampler_alpha": 1.0, |
| 93 | "use_language_weighted_sampler": false, |
| 94 | "language_weighted_sampler_alpha": 1.0, |
| 95 | "use_length_weighted_sampler": false, |
| 96 | "length_weighted_sampler_alpha": 1.0, |
| 97 | "model_args": { |
| 98 | "gpt_batch_size": 1, |
| 99 | "enable_redaction": false, |
| 100 | "kv_cache": true, |
| 101 | "gpt_checkpoint": null, |
| 102 | "clvp_checkpoint": null, |
| 103 | "decoder_checkpoint": null, |
| 104 | "num_chars": 255, |
| 105 | "tokenizer_file": "", |
| 106 | "gpt_max_audio_tokens": 605, |
| 107 | "gpt_max_text_tokens": 402, |
| 108 | "gpt_max_prompt_tokens": 70, |
| 109 | "gpt_layers": 30, |
| 110 | "gpt_n_model_channels": 1024, |
| 111 | "gpt_n_heads": 16, |
| 112 | "gpt_number_text_tokens": 6681, |
| 113 | "gpt_start_text_token": null, |
| 114 | "gpt_stop_text_token": null, |
| 115 | "gpt_num_audio_tokens": 1026, |
| 116 | "gpt_start_audio_token": 1024, |
| 117 | "gpt_stop_audio_token": 1025, |
| 118 | "gpt_code_stride_len": 1024, |
| 119 | "gpt_use_masking_gt_prompt_approach": true, |
| 120 | "gpt_use_perceiver_resampler": true, |
| 121 | "input_sample_rate": 22050, |
| 122 | "output_sample_rate": 24000, |
| 123 | "output_hop_length": 256, |
| 124 | "decoder_input_dim": 1024, |
| 125 | "d_vector_dim": 512, |
| 126 | "cond_d_vector_in_each_upsampling_layer": true, |
| 127 | "duration_const": 102400 |
| 128 | }, |
| 129 | "model_dir": null, |
| 130 | "languages": [ |
| 131 | "en", |
| 132 | "es", |
| 133 | "fr", |
| 134 | "de", |
| 135 | "it", |
| 136 | "pt", |
| 137 | "pl", |
| 138 | "tr", |
| 139 | "ru", |
| 140 | "nl", |
| 141 | "cs", |
| 142 | "ar", |
| 143 | "zh-cn", |
| 144 | "hu", |
| 145 | "ko", |
| 146 | "ja", |
| 147 | "hi" |
| 148 | ], |
| 149 | "temperature": 0.75, |
| 150 | "length_penalty": 1.0, |
| 151 | "repetition_penalty": 5.0, |
| 152 | "top_k": 50, |
| 153 | "top_p": 0.85, |
| 154 | "num_gpt_outputs": 1, |
| 155 | "gpt_cond_len": 30, |
| 156 | "gpt_cond_chunk_len": 4, |
| 157 | "max_ref_len": 30, |
| 158 | "sound_norm_refs": false |
| 159 | } |