HUMSANA SWAN

Voice AI Deepfake
Detection

Acoustic behavioral analysis across major TTS and RVC providers.

Works across PSTN, VoIP, and air-gap transmission paths.

Drop an audio file to analyze
WAV · MP3 · FLAC · OGG — auto-trimmed to 15 seconds
Analyzing...
Voice Authenticity Assessment
Synthetic Likelihood
Confidence
Signal Consensus

Results based on abbreviated analysis. Full-length evaluation available via API integration.

You've reached the free analysis limit. For additional analyses, API access starts at $49/month. Get in touch.

Multi Signal Domains

Each domain measures a distinct acoustic property. A determination requires consensus.

Spectral Variability
Measures fullband spectral flatness. Biological voice produces irregular spectral energy. Synthetic generation produces uniform distribution.
Cross-Band Energy
Compares energy distribution between low and mid frequency bands. Human laryngeal production creates natural divergence between bands.
Harmonic Phase Dispersion
Tracks phase coherence across harmonic partials. Physiological phonation produces natural phase drift that vocoders cannot replicate.
Speech Envelope Dynamics
Analyzes amplitude variation patterns across voiced segments. Biological speech exhibits higher dynamic range than synthesized output.
Pitch Micro-Modulation
Measures frame-to-frame F0 perturbation. Natural jitter from vocal fold biomechanics produces measurable micro-variation absent in synthesis.
Transition Morphology
Tracks phonation onset, offset, and post-burst behavior. Human speech transitions follow biomechanical trajectories that differ from generated audio.

Enterprise Integration

Deploy as a real-time control inside existing workflows.

REST API

Upload audio files for batch analysis. Returns determination, confidence, behavioral driver, and full forensic report.

POST /api/v1/analyze
Authorization: Bearer hv_...

Real-Time WebSocket

Stream live audio for in-call analysis. Receive determination events as they occur — sub-second adjudication.

WS /api/v1/stream?key=hv_...
→ {"event":"adjudication",...}

RTP Media Fork

Receive forked RTP streams directly from your SBC or media server. G.711 µ-law and A-law. Drop-in for contact center deployment.

UDP :5004 → Engine → Events

Audit Reports

Every determination produces a structured forensic report with full signal traceability. Dark-mode dashboards and print-ready audit exports.

GET /api/v1/sessions/{id}
→ Full session + report

Pricing

Start with 3 free analyses. Scale as you need.

Enterprise
Custom
Unlimited volume
Everything in API Access
RTP media fork integration
Custom policy mapping
Dedicated support + SLA
Contact Us

Ready to integrate?

Get API access and start analyzing voice interactions in production.

Contact Us