Upload Voice Audio
Upload a voice recording to start.
Upload a voice recording to start.
Local AI Voice Detector
Analyze AI-generated voice risk directly in your browser
Upload a voice recording and Resodock runs a DeepVoiceGuard ONNX model locally. The result is a probability-based screening score, not a legal proof of authenticity.
DeepVoiceGuard ONNXRuns locallyAudio is not uploaded
Upload Voice Audio
Supports wav/mp3/m4a/flac, max 100MB
Initializing upload...
Local model
Uses the DeepVoiceGuard RawNet ONNX model with 16kHz mono speech windows. The model is cached after first download.
The audio file is decoded and analyzed in the browser. The model is fetched from the CDN, not the audio.
Upload a voice recording to start.
A sponsor page opens in a new tab while processing continues here.
When an AI voice detector is useful
AI voice detection is most useful as an early screening step for suspicious voice notes, social clips, interview audio, phone recordings, and creator submissions.
The result includes a probability score and segment-level timeline so you can see which parts of the file pushed the score higher.
Why local detection matters
Voice recordings can include private conversations, names, addresses, or work material. Keeping the analysis in the browser avoids sending the audio file to a server just to get a first check.
The first model load is larger, but IndexedDB caching means repeat checks can start much faster on the same device.
AI Voice Detector FAQ
Can this prove a voice is AI-generated?
No. It provides a probability-based risk score. Treat it as a screening tool, especially when audio has been compressed or edited.
Is my audio uploaded?
No. The audio is decoded and analyzed in the browser. The model file is downloaded and cached locally.
What model does it use?
It uses the DeepVoiceGuard RawNet ONNX model trained for synthetic speech detection.