GPT-SoVITS模型|小顽声音模型小顽AI配音模型

擅长领域：童声/情感/动漫/游戏｜可爱男孩，自然动听

GPT-SoVITS模型配音效果

鉴于GPT-SoVITS模型自回归特性，即其配音情绪高度依赖于所提供的参考音频，特此说明：本视频所展示的配音情绪仅为采用某一特定参考音频时的效果示例，并不全面反映GPT-SoVITS模型能够生成的全部情绪范围及最终配音质量的上限。模型的最终表现将随着不同参考音频的输入而展现出多样化。

训练日志

2025-07-31 15:23:24,851	魔小顽	INFO	{'train': {'log_interval': 100, 'eval_interval': 500, 'seed': 1234, 'epochs': 20, 'learning_rate': 0.0001, 'betas': [0.8, 0.99], 'eps': 1e-09, 'batch_size': 11, 'fp16_run': True, 'lr_decay': 0.999875, 'segment_size': 20480, 'init_lr_ratio': 1, 'warmup_epochs': 0, 'c_mel': 45, 'c_kl': 1.0, 'text_low_lr_rate': 0.4, 'pretrained_s2G': 'GPT_SoVITS/pretrained_models/s2G488k.pth', 'pretrained_s2D': 'GPT_SoVITS/pretrained_models/s2D488k.pth', 'if_save_latest': True, 'if_save_every_weights': True, 'save_every_epoch': 20, 'gpu_numbers': '0'}, 'data': {'max_wav_value': 32768.0, 'sampling_rate': 32000, 'filter_length': 2048, 'hop_length': 640, 'win_length': 2048, 'n_mel_channels': 128, 'mel_fmin': 0.0, 'mel_fmax': None, 'add_blank': True, 'n_speakers': 300, 'cleaned_text': True, 'exp_dir': 'logs/魔小顽'}, 'model': {'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': [10, 8, 2, 2, 2], 'upsample_initial_channel': 512, 'upsample_kernel_sizes': [16, 16, 8, 2, 2], 'n_layers_q': 3, 'use_spectral_norm': False, 'gin_channels': 512, 'semantic_frame_rate': '25hz', 'freeze_quantizer': True}, 's2_ckpt_dir': 'logs/魔小顽', 'content_module': 'cnhubert', 'save_weight_dir': 'SoVITS_weights', 'name': '魔小顽', 'pretrain': None, 'resume_step': None}
2025-07-31 15:23:26,112	魔小顽	INFO	loaded pretrained GPT_SoVITS/pretrained_models/s2G488k.pth
2025-07-31 15:23:26,329	魔小顽	INFO	loaded pretrained GPT_SoVITS/pretrained_models/s2D488k.pth
2025-07-31 15:23:54,552	魔小顽	INFO	Train Epoch: 1 [0%]
2025-07-31 15:23:54,553	魔小顽	INFO	[2.4472553730010986, 2.469399929046631, 11.060148239135742, 22.705406188964844, 0.0, 2.0762078762054443, 0, 9.99875e-05]
2025-07-31 15:24:08,779	魔小顽	INFO	====> Epoch: 1
2025-07-31 15:24:23,819	魔小顽	INFO	====> Epoch: 2
2025-07-31 15:24:37,848	魔小顽	INFO	====> Epoch: 3
2025-07-31 15:24:52,040	魔小顽	INFO	====> Epoch: 4
2025-07-31 15:25:07,455	魔小顽	INFO	====> Epoch: 5
2025-07-31 15:25:22,392	魔小顽	INFO	====> Epoch: 6
2025-07-31 15:25:29,553	魔小顽	INFO	Train Epoch: 7 [25%]
2025-07-31 15:25:29,554	魔小顽	INFO	[2.38828706741333, 2.5152785778045654, 10.330231666564941, 18.805980682373047, 0.0, 1.966209053993225, 100, 9.991253280566489e-05]
2025-07-31 15:25:37,635	魔小顽	INFO	====> Epoch: 7
2025-07-31 15:25:52,052	魔小顽	INFO	====> Epoch: 8
2025-07-31 15:26:06,150	魔小顽	INFO	====> Epoch: 9
2025-07-31 15:26:21,345	魔小顽	INFO	====> Epoch: 10
2025-07-31 15:26:36,109	魔小顽	INFO	====> Epoch: 11
2025-07-31 15:26:50,828	魔小顽	INFO	====> Epoch: 12
2025-07-31 15:26:59,933	魔小顽	INFO	Train Epoch: 13 [50%]
2025-07-31 15:26:59,934	魔小顽	INFO	[2.242635726928711, 2.8274648189544678, 11.853876113891602, 20.44843292236328, 0.0, 1.6038278341293335, 200, 9.983762181915804e-05]
2025-07-31 15:27:05,697	魔小顽	INFO	====> Epoch: 13
2025-07-31 15:27:19,866	魔小顽	INFO	====> Epoch: 14
2025-07-31 15:27:34,804	魔小顽	INFO	====> Epoch: 15
2025-07-31 15:27:49,134	魔小顽	INFO	====> Epoch: 16
2025-07-31 15:28:04,434	魔小顽	INFO	====> Epoch: 17
2025-07-31 15:28:18,971	魔小顽	INFO	====> Epoch: 18
2025-07-31 15:28:31,908	魔小顽	INFO	Train Epoch: 19 [75%]
2025-07-31 15:28:31,909	魔小顽	INFO	[2.1451520919799805, 2.9453847408294678, 11.223749160766602, 19.860918045043945, 0.0, 1.4945650100708008, 300, 9.976276699833672e-05]
2025-07-31 15:28:34,365	魔小顽	INFO	====> Epoch: 19
2025-07-31 15:28:49,183	魔小顽	INFO	Saving model and optimizer state at iteration 20 to logs/魔小顽/logs_s2\G_233333333333.pth
2025-07-31 15:28:49,970	魔小顽	INFO	Saving model and optimizer state at iteration 20 to logs/魔小顽/logs_s2\D_233333333333.pth
2025-07-31 15:28:52,030	魔小顽	INFO	saving ckpt 魔小顽_e20:Success.
2025-07-31 15:28:52,031	魔小顽	INFO	====> Epoch: 20