擅长领域： 稚嫩，可爱，童真无邪，童声，动漫，游戏
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
2024-07-31 21:26:16,211 魔小洛 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/魔小洛'}, '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/魔小洛', 'content_module': 'cnhubert', 'save_weight_dir': 'SoVITS_weights', 'name': '魔小洛', 'pretrain': None, 'resume_step': None}
2024-07-31 21:26:17,627 魔小洛 INFO loaded pretrained GPT_SoVITS/pretrained_models/s2G488k.pth
2024-07-31 21:26:17,854 魔小洛 INFO loaded pretrained GPT_SoVITS/pretrained_models/s2D488k.pth
2024-07-31 21:26:51,698 魔小洛 INFO Train Epoch: 1 [0%]
2024-07-31 21:26:51,698 魔小洛 INFO [2.6779346466064453, 2.520355701446533, 8.829712867736816, 21.966920852661133, 0.0, 2.3901848793029785, 0, 9.99875e-05]
2024-07-31 21:27:09,188 魔小洛 INFO ====> Epoch: 1
2024-07-31 21:27:26,097 魔小洛 INFO ====> Epoch: 2
2024-07-31 21:27:42,622 魔小洛 INFO ====> Epoch: 3
2024-07-31 21:27:58,716 魔小洛 INFO ====> Epoch: 4
2024-07-31 21:28:15,604 魔小洛 INFO ====> Epoch: 5
2024-07-31 21:28:22,943 魔小洛 INFO Train Epoch: 6 [26%]
2024-07-31 21:28:22,944 魔小洛 INFO [2.4876067638397217, 2.570812225341797, 8.132889747619629, 18.708391189575195, 0.0, 1.814195990562439, 100, 9.99250234335941e-05]
2024-07-31 21:28:33,097 魔小洛 INFO ====> Epoch: 6
2024-07-31 21:28:49,429 魔小洛 INFO ====> Epoch: 7
2024-07-31 21:29:07,106 魔小洛 INFO ====> Epoch: 8
2024-07-31 21:29:22,464 魔小洛 INFO ====> Epoch: 9
2024-07-31 21:29:38,622 魔小洛 INFO ====> Epoch: 10
2024-07-31 21:29:49,375 魔小洛 INFO Train Epoch: 11 [53%]
2024-07-31 21:29:49,376 魔小洛 INFO [2.405259609222412, 2.314974308013916, 7.7199225425720215, 17.42384910583496, 0.0, 1.7544214725494385, 200, 9.986258590528146e-05]
2024-07-31 21:29:55,375 魔小洛 INFO ====> Epoch: 11
2024-07-31 21:30:11,348 魔小洛 INFO ====> Epoch: 12
2024-07-31 21:30:27,682 魔小洛 INFO ====> Epoch: 13
2024-07-31 21:30:43,780 魔小洛 INFO ====> Epoch: 14
2024-07-31 21:30:59,383 魔小洛 INFO ====> Epoch: 15
2024-07-31 21:31:12,792 魔小洛 INFO Train Epoch: 16 [79%]
2024-07-31 21:31:12,793 魔小洛 INFO [2.3955600261688232, 2.536201238632202, 8.378169059753418, 18.136430740356445, 0.0, 1.7185262441635132, 300, 9.980018739066937e-05]
2024-07-31 21:31:15,201 魔小洛 INFO ====> Epoch: 16
2024-07-31 21:31:31,317 魔小洛 INFO ====> Epoch: 17
2024-07-31 21:31:47,552 魔小洛 INFO ====> Epoch: 18
2024-07-31 21:32:03,420 魔小洛 INFO ====> Epoch: 19
2024-07-31 21:32:18,918 魔小洛 INFO Saving model and optimizer state at iteration 20 to logs/魔小洛/logs_s2\G_233333333333.pth
2024-07-31 21:32:19,709 魔小洛 INFO Saving model and optimizer state at iteration 20 to logs/魔小洛/logs_s2\D_233333333333.pth
2024-07-31 21:32:21,254 魔小洛 INFO saving ckpt 魔小洛_e20:Success.
2024-07-31 21:32:21,254 魔小洛 INFO ====> Epoch: 20如何使用配音模型
1，gpt-sovits模型云端部署
https://aiaf.cc/gpt-sovits-yunduan/.html
2，gpt-sovits模型本地部署
https://aiaf.cc/gpt-sovits/.html
如果您想一对一远程教学模型安装、模型训练，请联系微信 xiaoming1870
声音版权使用声明
本网站展示的 AI 声音模型由站长及工作室精心创作并提供。遵循非商业性使用原则，仅作娱乐用途，重视并遵守版权所有者权益，未获授权也不声称拥有使用权。模型整理等产生的费用仅覆盖服务成本，不涉及版权收费。所有活动在法律框架内进行，尊重版权、合法使用分享。如有疑问、需版权信息或建议反馈，可随时联系，共同促进 AI 声音艺术发展与营造尊重版权氛围。
没有回复内容