GPT-SoVITS UVR5 OS Command Injection in Output Paths Allows Unauthenticated Server Code Execution

The GPT-SoVITS voice cloning application contains a critical OS command injection vulnerability within its UVR5 module. According to the National Vulnerability Database record for CVE-2025-71429, this flaw permits unauthenticated attackers to execute arbitrary commands on the host system. The issue resides in the tools/uvr5/vr.py script, which processes user input without adequate sanitisation.

Root cause

The vulnerability exists in the tools/uvr5/vr.py file. According to the NVD description, the application allows attackers to execute commands via output folder paths. The flaw enables the injection of shell substitutions, such as $(...), into the processing logic. This mechanism allows the operating system to interpret user-supplied path data as executable instructions.

Attack path

An attacker interacts with the UVR5 WebUI interface to trigger the exploit. They specify a malicious string in the output folder textbox when selecting MP3 or M4A export formats. The application then passes this input to the underlying shell. The server executes the injected commands with the privileges of the user running the GPT-SoVITS service, granting control over that account.

Affected versions

• GPT-SoVITS versions through 20250606v2pro are affected.
• The vulnerability is present in the RVC-Boss maintained fork of the project.
• The NVD record identifies the specific version range for CVE-2025-71429.

Mitigation

• Avoid exposing the UVR5 WebUI to untrusted networks or the public internet.
• Restrict access to the application using strict firewall rules or network segmentation.
• Monitor server logs for unusual shell activity or unexpected process executions.
• Isolate the GPT-SoVITS service in a container with minimal privileges and no access to critical system files.

Background: National Vulnerability Database

The National Vulnerability Database provides standard identifiers for software flaws. You use these identifiers to automate patching and verify vendor claims. Small teams should cross-reference database entries with their IT asset inventory to prioritise fixes and avoid wasting time on irrelevant alerts.

Read the full guide: National Vulnerability Database: Small Business Action Plan

What to do and how to stay safe: GPT-SoVITS

• Audit your deployment to ensure the UVR5 module is not accessible from untrusted sources.
• Implement network-level access controls to limit who can reach the WebUI interface.
• Monitor system logs for signs of command execution originating from the audio processing service.
• Once the vendor provides an update, apply it immediately to remediate the injection flaw.

Step-by-step guide: Security Advisories: 6 Common Misreadings That Leave Systems Exposed

General security guidance from the Vector Bulletin newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Which specific component is vulnerable in GPT-SoVITS?

The vulnerability is located in the tools/uvr5/vr.py file within the UVR5 module.

Can this flaw be exploited remotely?

Yes, unauthenticated attackers can exploit this via the WebUI if it is accessible.

Is there a patch available for CVE-2025-71429?

No fix has been confirmed yet; users should apply network mitigations until a patch is released.