GPT-SoVITS
GPT-SoVITS copied to clipboard
RVC-Boss
1 min voice data can also be used to train a good TTS model! (few shot voice cloning)
在跑訓練GPT時遇到的問題 "C:\Users\rober\OneDrive\桌面\GPT-SoVITS\runtime\python.exe" GPT_SoVITS/s1_train.py --config_file "TEMP/tmp_s1.yaml" Seed set to 1234 Using 16bit Automatic Mixed Precision (AMP) GPU available: True (cuda), used: True TPU available: False, using: 0 TPU cores IPU available:...
一键三连报错
"D:\software\GPT-SoVITS-beta\GPT-SoVITS-beta0128\runtime\python.exe" GPT_SoVITS/prepare_datasets/1-get-text.py "D:\software\GPT-SoVITS-beta\GPT-SoVITS-beta0128\runtime\python.exe" GPT_SoVITS/prepare_datasets/1-get-text.py Traceback (most recent call last): Traceback (most recent call last): File "D:\software\GPT-SoVITS-beta\GPT-SoVITS-beta0128\GPT_SoVITS\prepare_datasets\1-get-text.py", line 10, in File "D:\software\GPT-SoVITS-beta\GPT-SoVITS-beta0128\GPT_SoVITS\prepare_datasets\1-get-text.py", line 10, in os.environ["CUDA_VISIBLE_DEVICES"] = os.environ.get("_CUDA_VISIBLE_DEVICES") os.environ["CUDA_VISIBLE_DEVICES"] = os.environ.get("_CUDA_VISIBLE_DEVICES")...
The described model architecture is same as the XTTS. To be fair, you should give a reference to their work as I stated #156
已经 brew install ffmpeg + conda install ffmpeg 过了
ffmpeg 虽然在第一步切割文件slice中已经可以正常工作,但是在第二个tab一件三连中再次报错，更具体的说时在SSL提取时。 Traceback (most recent call last): File "D:\dev\replica\GPT-SoVITS\GPT_SoVITS\prepare_datasets\2-get-hubert-wav32k.py", line 103, in name2go(wav_name) File "D:\dev\replica\GPT-SoVITS\GPT_SoVITS\prepare_datasets\2-get-hubert-wav32k.py", line 68, in name2go tmp_audio = load_audio(wav_path, 32000) File "D:\dev\replica\GPT-SoVITS\tools\my_utils.py", line 20, in load_audio raise...
我是在服务器上运行的 "/home/jingyc/.conda/envs/gptsovits/bin/python" GPT_SoVITS/prepare_datasets/1-get-text.py "/home/jingyc/.conda/envs/gptsovits/bin/python" GPT_SoVITS/prepare_datasets/1-get-text.py Traceback (most recent call last): File "/home/jingyc/.conda/envs/gptsovits/lib/python3.9/site-packages/transformers/modeling_utils.py", line 533, in load_state_dict return torch.load( File "/home/jingyc/.conda/envs/gptsovits/lib/python3.9/site-packages/torch/serialization.py", line 814, in load raise pickle.UnpicklingError(UNSAFE_MESSAGE + str(e)) from None...
# 1.28.2024 在原api.py基础上做出的一些改动 ## 简介 - 原接口不变，仿照silero-api-server格式添加了一些endpoint，可接入傻酒馆sillytavern。 - 运行api.py直至显示http://127.0.0.1:9880 - 在staging版本的sillytavern>Extensions>TTS>Select TTS Provider选择silero - 将http://127.0.0.1:9880填入Provider Endpoint后点击reload - Select TTS Provider上方显示TTS Provider Loaded则连接成功，之后照常设置即可。 - 支持运行中根据讲话人名称自动更换声音模型或参考音频。 - 如果运行api.py时使用-vd提供了声音模型根目录，可以根据讲话人名称（子文件夹名称或"default"）自动更换模型和参考音频。例如： python api.py -vd "D:/Voices"...
