Human or AI?
Your audio
English · 1–60 secAudio preview isn’t available in this browser.
You can still run detection.
WAV, MP3, FLAC or OGG · up to 12 MiB
Choose a clip to get started.
Audio isn’t stored or used for training.
The signal
Model prediction
A model score, not a guarantee of authenticity.
View model response
How this AI speech detector works
Upload a 1–60 second speech clip, then press Analyze audio. OrukZero looks for patterns associated with human and synthetic speech. Longer clips include scores for individual segments.
What does the AI score mean?
The score is a model signal, not a calibrated probability or proof of authenticity. A human result does not guarantee a real recording; an AI result does not establish fraud or identify the voice generator. Short, noisy, edited or unfamiliar audio can produce mistakes. This detector is for speech, not music.
What happens to my audio?
The demo uploads audio only after you press Analyze. Application audio is processed transiently, not retained or used for training. The public demo has usage limits; operational telemetry is described in our privacy policy.
Can I use the detection API?
OrukZero’s synthetic speech detection API is included at no extra charge with every active paid Oruk plan. Read the API documentation.