AI Voice Detector

About

AI Voice Detector tells you whether the voice you're listening to in a browser tab is a real human or an AI-generated synthetic voice. It works on any tab that plays audio — video calls, news clips, podcasts, social posts — and produces a verdict in real time.

How it works
• Click the toolbar icon and choose "Start" on the active tab.
• The extension captures the tab's audio stream locally.
• An on-device machine-learning model analyzes the voice and produces a verdict: human, AI-generated, or probably AI-generated.
• A small overlay in the page shows the live verdict. Stop at any time from the popup.

Everything happens on your device
• No audio ever leaves your computer.
• No recordings are saved.
• No external server receives the audio, inference results, or page content.
• The ML models are bundled inside the extension and run locally in your browser.

What you need
• Chrome (or a Chromium-based browser) with extension support.
• A tab that is currently playing audio.

Permissions, in plain language
• tabCapture — to read the audio from the tab you explicitly ask to analyze.
• offscreen — to run audio decoding and ML inference without freezing the browser.
• activeTab + scripting — to show the verdict overlay on the tab you're analyzing.
• storage — to remember anonymous usage-analytics identifiers for the popup UI.

Important notice — please read

By using the AI Voice Detector, you consent to the real-time analysis of audio played within your browser tab. The tool processes audio locally on your device to determine whether the voice is human or AI-generated. No audio data is stored, recorded, or transmitted to any external servers — all processing occurs entirely on your device.

The AI Voice Detector does not process or analyze biometric data for identification or profiling purposes. It solely analyzes audio to distinguish between human and AI-generated voices. No personal or biometric data is collected, stored, or used.

The AI Voice Detector provides analysis based on an algorithm that classifies audio as human-generated, AI-generated, or probably AI-generated. While designed to be accurate, the tool may not be 100% precise in all cases, particularly in complex or mixed audio environments. Results should be treated as indicative rather than definitive, and users are encouraged to consider the tool's limitations when interpreting the analysis.