GPT-SoVITS UVR5 Command Injection: Unauthenticated Shell Code in Output Fields Yields CVSS 9.3 RCE

RVC-Boss GPT-SoVITS contains a critical OS command injection vulnerability tracked as CVE-2025-71429. The flaw allows unauthenticated attackers to execute arbitrary system commands by manipulating output folder paths in the UVR5 WebUI.

What happened

The vulnerability exists in the tools/uvr5/vr.py script within GPT-SoVITS versions through 20250606v2pro. According to the NVD record, attackers can inject shell substitutions such as $(...) into the output folder textboxes. This occurs specifically when users export audio files in mp3 or m4a formats.

The source material states that these actions allow attackers to run commands as the server user. The National Vulnerability Database assigns a CVSS score of 9.3, categorizing the severity as critical.

The weakness is identified as CWE-78, which relates to improper neutralization of special elements used in an OS command. This classification highlights the direct execution risk posed by the vulnerability in the specified component.

Why it matters

Developers deploying GPT-SoVITS must understand that the UVR5 WebUI is vulnerable to this specific injection technique. Any user with access to the web interface can potentially exploit the flaw without authentication.

The unauthenticated nature of the attack vector significantly increases the risk profile. Attackers can execute commands as the server user, potentially compromising the integrity of the service.

The affected component, vr.py, is integral to the voice conversion and restoration workflow. This means the vulnerability is present in a core feature used by many voice cloning and audio processing workflows.

What to watch

• Monitor GPT-SoVITS repositories for updates addressing CVE-2025-71429.
• Review server logs for unexpected command executions originating from the UVR5 module.
• Check for unauthorized changes to system files or new user accounts on affected servers.

Background: CVSS score

CVSS provides a standardized vector for exploit difficulty and access requirements, not business value. A high score indicates an easy attack path, not necessarily significant damage. You must contextualize the score with your environment’s exposure, asset value, and compensating controls to determine true risk.

Read the full guide: CVSS Scores: Why High Numbers Mislead Your Risk Assessment

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

• Restrict network access to the GPT-SoVITS WebUI to trusted IP addresses only.
• Disable the UVR5 module if it is not currently needed for your workflow.
• Ensure the server user account has limited permissions to minimize potential damage.
• Monitor system processes for unusual activity that may indicate command injection attempts.

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

Frequently asked questions

Which GPT-SoVITS versions are vulnerable to CVE-2025-71429?

All versions of GPT-SoVITS up to and including 20250606v2pro are affected by this vulnerability.

How do attackers exploit this command injection flaw?

Attackers enter shell substitutions like $(...) into the output folder textboxes when exporting mp3 or m4a files in the UVR5 WebUI.

What is the CVSS score for CVE-2025-71429?

The National Vulnerability Database rates the severity of CVE-2025-71429 as critical with a CVSS score of 9.3.