Assess AI voice authenticity signals without identifying the speaker.

Intended audience

• creators
• voice agencies
• media teams
• internal trust teams
• authorized individual users

Primary job

Assess whether an audio file shows signals associated with AI synthesis, voice conversion, editing, replay, re-recording, or provenance anomalies.

Data mode

ephemeral_local_audio_signal_mode

• one Human-selected local audio file
• synthetic test fixtures
• approved benchmark fixtures

Primary output

Voice Authenticity Signal Assessment

No strong anomaly signal does not prove that the audio is authentic and does not prove speaker identity.

Prohibited capabilities

• local execution only
• no server upload
• no server persistence
• no training use
• no microphone capture
• no remote URL fetch
• no speaker identification
• no voiceprint generation
• no speaker embedding persistence
• no biometric identity decision

Fixed limitation

This surface does not identify the speaker, prove that audio is authentic, prove that audio is fake, prove ownership, determine infringement, determine fraud, or provide a legal conclusion.

Runtime status: SURFACE_DEFINED_ENGINE_NOT_ACTIVE