A flaw has been found in RVC-Project Retrieval-based-Voice-Conversion-WebUI up to 2.3.260718. Impacted is the function changeinfo of the file webui.py. Executing a manipulation of the argument ckptpath can lead to code injection. The attack can be executed remotely. The exploit has been published and may be used. The project was informed of the problem early through an issue report but has not responded yet.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to command injection. The variables expdir1, np7, trainsetdir4 and sr2 take user input and pass it to the preprocessdataset function, which concatenates them into a command that is run on the server. This can lead to arbitrary command execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to command injection. The variables expdir1, np7 and f0method8 take user input and pass it into the extractf0feature function, which concatenates them into a command that is run on the server. This can lead to arbitrary command execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to command injection. The variables expdir1, among others, take user input and pass it to the clicktrain function, which concatenates them into a command that is run on the server. This can lead to arbitrary command execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to unsafe deserialization. The ckptpath1 variable takes user input (e.g. a path to a model) and passes it to the showinfo function in processckpt.py, which uses it to load the model on that path with torch.load, which can lead to unsafe deserialization and remote code execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to code injection. The ckptpath2 variable takes user input (e.g. a path to a model) and passes it to changeinfo function, which opens and reads the file on the given path (except it changes the final on the path to train.log), and passes the contents of the file to eval, which can lead to remote code execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to unsafe deserialization. The ckptpath2 variable takes user input (e.g. a path to a model) and passes it to the extractsmallmodel function in processckpt.py, which uses it to load the model on that path with torch.load, which can lead to unsafe deserialization and remote code execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to unsafe deserialization. The ckptpath0 variable takes user input (e.g. a path to a model) and passes it to the changeinfo function in processckpt.py, which uses it to load the model on that path with torch.load, which can lead to unsafe deserialization and remote code execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to unsafe deserialization. The ckptdir variable takes user input (e.g. a path to a model) and passes it to the changeinfo function in export.py, which uses it to load the model on that path with torch.load, which can lead to unsafe deserialization and remote code execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to unsafe deserialization. The modelchoose variable takes user input (e.g. a path to a model) and passes it to the uvr function in vr.py. In uvr , if modelname contains the string "DeEcho", a new instance of AudioPreDeEcho class is created with the modelpath attribute containing the aforementioned user input. In the AudioPreDeEcho class, the user input is used to load the model on that path with torch.load, which can lead to unsafe deserialization and remote code execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to unsafe deserialization. The ckpta and cpktb variables take user input (e.g. a path to a model) and pass it to the merge function in processckpt.py, which uses them to load the models on those paths with torch.load, which can lead to unsafe deserialization and remote code execution. As of time of publication, no known patches exist.

Retrieval-based-Voice-Conversion-WebUI is a voice changing framework based on VITS. Versions 2.2.231006 and prior are vulnerable to unsafe deserialization. The modelchoose variable takes user input (e.g. a path to a model) and passes it to the uvr function in vr.py. In uvr , a new instance of AudioPre class is created with the modelpath attribute containing the aformentioned user input. In the AudioPre class, the user input, is used to load the model on that path with torch.load, which can lead to unsafe deserialization and remote code execution. As of time of publication, no known patches exist.