Organisations at Risk: GPT-SoVITS UVR5 Flaw Enables Unauthenticated System Command Execution

RVC-Boss GPT-SoVITS versions up to 20250606v2pro are affected by a critical security flaw. According to the National Vulnerability Database, the issue is an OS command injection vulnerability located in the UVR5 module. The NVD record assigns CVE-2025-71429 a CVSS score of 9.3. This score classifies the weakness as Critical. The vulnerability exists within the tools/uvr5/vr.py file of the application. No patched version is mentioned in the available source material.

How it unfolded

• The NVD published a record for CVE-2025-71429 detailing the flaw in GPT-SoVITS.
• The record identifies CWE-78 as the underlying weakness type for the vulnerability.
• The description specifies that the issue allows command execution via output folder paths.

Who is affected

Organizations using GPT-SoVITS up to version 20250606v2pro are at risk. The National Vulnerability Database states the flaw allows unauthenticated attackers to execute commands. The attack vector involves the UVR5 WebUI output folder textboxes. Attackers can input shell substitutions such as $(...) when exporting mp3 or m4a files. This action runs commands as the server user. The vulnerability is inherent in the handling of file paths within the specified versions.

Our take on the response

The source material provides technical details but does not confirm a vendor release. That suggests administrators must verify their current version against the affected list immediately. The lack of a confirmed patch in the record means exposure remains high for all unpatched instances. Teams should treat this as an urgent matter requiring immediate inventory checks.

Who needs to act first

Managed service providers should prioritize systems exposed to the internet. Check whether GPT-SoVITS instances are accessible without strong network segmentation. Teams usually begin by isolating affected servers from untrusted networks. Verify if the UVR5 module is enabled on public-facing endpoints. Reduce exposure by restricting access to the web interface. Ensure backup procedures are current before making configuration changes. Monitor logs for unusual command execution patterns. Coordinate with internal security teams to validate the current software versions.

Background: Attack vectors

Attack vectors are the paths adversaries use to reach your data. They range from simple phishing emails to complex supply chain compromises. Defending against them requires mapping every entry point, validating inputs, and limiting what each component can do. You cannot stop all attacks, but you can make them costly and slow.

Read the full guide: Attack Vectors Explained: Mechanisms, Limits, and Defense Strategies

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

• Verify if your deployment is running GPT-SoVITS version 20250606v2pro or earlier.
• Restrict network access to the UVR5 WebUI to trusted internal IPs only.
• Monitor server logs for unexpected command execution or shell substitution attempts.
• Isolate affected systems from the internet until a vendor update is confirmed.

Step-by-step guide: Vulnerability Prioritization: How to Rank Risks Without Wasting Time

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

Frequently asked questions

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

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

Which versions of GPT-SoVITS are affected?

Versions of GPT-SoVITS up to and including 20250606v2pro are affected.

How can attackers exploit this vulnerability?

Attackers can enter shell substitutions in the UVR5 WebUI output folder textboxes.