SmrAI Logo
NEURAL ACOUSTIC FORENSICS V7.1 • MULTI-LAYER SPEECH DEEPFAKE SUITE speed Latency: 24ms • 96kHz / 32-bit Float verified Models: ElevenLabs v2, OpenAI Voice, XTTS, Suno, C2PA Audio

AI Voice Detector Free & Deepfake Audio Analyzer

Free AI voice detector and acoustic deepfake analyzer. Detect ElevenLabs cloned voices, synthetic speech, TTS artifacts, and biometrics anomalies online with zero data retention.

graphic_eq Preset:
layers Layer:
Master Acoustic Composite
Dual-Domain View: Waveform + Mel-Spectrogram (0Hz – 24kHz)
Time-Domain Amplitude Waveform (RMS Envelope) -12.4 LUFS
Mel-Spectrogram (0Hz – 24kHz) • Formant Harmonics warning HiFi-GAN 16kHz Cutoff
24.0 kHz 18.0 kHz 16.0 kHz (Cutoff) 8.0 kHz 0 Hz
00:04.12 / 00:14.80
Bitrate: 96kHz / 32-bit Float PCM • Channels: Stereo (2.0) • Codec: Lossless WAV
Vocoder Phase Discontinuity Detected
layers

Master Acoustic Composite (All 5 Forensic Layers)

Full-Spectrum Cross-Correlated Speech Forensics

SYNTHETIC VOCODER ANOMALY DETECTED

Complete acoustic overview merging Mel-frequency spectral distribution, glottal micro-jitter, amplitude shimmer, neural vocoder subharmonics, and biometric vocal tract geometry.

Glottal Jitter
0.01% (Flatline)
Violates vocal fold biomechanics
Spectral Cutoff
16.0 kHz Brickwall
HiFi-GAN vocoder truncation
Phase Notch
4.2 kHz Discontinuity
Phoneme concatenation artifact
Deepfake Countermeasure: Cross-domain higher-order spectral decomposition Confidence: 96.8%
mic

Audio & Hardware Dossier

VIRTUAL DSP BUFFER
Origin: Neural Vocoder TTS
mic Acoustic Hardware Full detail info
None (Virtual Server)
ElevenLabs Cluster (HiFi-GAN)
Transducer: Latent Vocoder DSP
mic Capture Hardware Dossier
FULL DETAIL
Microphone Hardware Model
None (Virtual Server)
DSP Engine / Manufacturer
ElevenLabs Cluster (HiFi-GAN)
Transducer & Diaphragm
Transducer: Latent Vocoder DSP
Hardware Pipeline: cancel Synthetic Buffer
Click card to inspect full stream metadata open_in_new
calendar_month Stream Identity Full detail info
October 11, 2026
elevenlabs_adam_v2.wav
Specs: 96kHz • 32-bit Float • 14.8s
calendar_month Audio Stream Identity
FULL DETAIL
Date & Time Recorded
October 11, 2026, 06:14:22 UTC
File Name & Container
elevenlabs_adam_v2.wav (RIFF PCM)
Audio Specifications
Specs: 96kHz • 32-bit Float • 14.8s • Stereo
Ingestion SHA-256: Verified Header
Click card to inspect RIFF headers open_in_new
vital_signs Acoustic Dynamics Full detail info
Jitter: 0.01% (Flatline)
Shimmer: 0.03 dB • RMS -12.4
HNR: 31.4 dB (Sterile Noise Floor)
vital_signs Acoustic Dynamics & Jitter
FULL DETAIL
Glottal Pitch Jitter (PPQ)
0.01% (Biological threshold: 0.5%–1.2%)
Amplitude Shimmer & RMS
Shimmer: 0.03 dB • RMS: -12.4 LUFS
Harmonics-to-Noise Ratio (HNR)
HNR: 31.4 dB (Unnaturally Sterile Floor)
Respiration Noise Floor: warning Absent Chest Intake
Click card to inspect F0 pitch curve open_in_new
verified C2PA Provenance Full detail info
AI Watermark Detected
SynthID / ElevenLabs Watermark
Cryptoseal: SHA-256 Ledger Anchored
verified_user C2PA Acoustic Cryptoseal
FULL DETAIL
C2PA Manifest Status
AI Watermark Signature Detected
Watermarking Signature
ElevenLabs Encrypted Audio Watermark
Legal Chain of Custody
Cryptographic SHA-256 Ledger Anchored
Admissibility Status: gavel Legally Certified Audit
Click card to inspect certificate open_in_new
Forensic Speech Verdict CONFIRMED SYNTHETIC
96.8% Cloned Audio

Matches ElevenLabs voice synthesis profiles with unnatural phase synchronization and vocoder artifacts.

Acoustic Model Signature Multi-Model Classification
ElevenLabs Multilingual v2 94.2%
OpenAI Advanced Voice 3.8%
XTTS v2 / Bark Generative 1.5%
Biological Human Respiration 0.5%
Forensic Diagnostics Breakdown
vital_signs
Biological Breath Micro-tremor Flatline (0.01% organic variation)

Absent biological pulse dynamics and vocal fold tremor in speech pauses.

view_timeline
Phoneme Glottal Symmetry Synthetic Perfect (Artificial Glottis)

Vocal tract geometry lacks natural asymmetric pitch perturbations.

waves
Formant Phase Continuity Discontinuity at 4.2 kHz

Neural vocoder concatenation artifacts detected at phoneme boundaries.

Real-Time Call Stream Defense

SDK v2.1.4 • C / WebRTC / Python

Intercept synthetic impersonators and unauthorized AI voice agents inside live telecommunication streams, Twilio endpoints, and enterprise VoIP switches with sub-80ms inline mitigation.

> npm i @smrai/stream-voice-guard
Audio Trust Standards
shield
SOC2 Type II Certified Continuous acoustic compliance auditing
lock
ISO 27001 Audio Vault AES-256 encrypted ingest buffer
delete_forever
Zero Voice Retention Instant cryptographic flush after scan
graphic_eq Acoustic Forensics

How Our AI Voice Detector Exposes Audio Deepfakes

Modern voice cloners like ElevenLabs can replicate timbre in minutes, but they leave telltale acoustic footprints. SmrAI is the premier ai voice detector free online, inspecting audio across five distinct physical and spectral biometric dimensions.

waves

48 kHz Mel-Spectrogram

High-resolution spectral analysis identifies unnatural 16.0 kHz brickwall frequency roll-offs and phase discontinuities characteristic of neural vocoders.

vital_signs

F0 Pitch Jitter & Breath Dynamics

Real human vocal cords exhibit organic micro-tremor and sub-glottal pressure shifts. Cloned voices produce unnaturally uniform fundamental frequencies.

mic_external_on

Vocoder Attribution Engine

Our deepfake audio detector isolates model-specific neural signatures, attributing samples to ElevenLabs, XTTS, OpenAI Voice, or Suno with acoustic forensic precision.

help_outline Common Queries

Frequently Asked Questions: AI Voice Detector

Everything you need to know about detecting voice clones and deepfake speech.

How does the free AI voice detector work? expand_more

SmrAI inspects high-resolution 48 kHz Mel-spectrograms for neural vocoder signatures, phase discrepancies, unnatural pitch stability (lack of biological jitter), and artificial 16 kHz frequency brickwalls.

Can SmrAI detect ElevenLabs, OpenAI Voice, and Suno deepfakes? expand_more

Yes. Our models are trained on all major speech synthesis architectures including ElevenLabs v1/v2, OpenAI Voice Engine, XTTS, Bark, Tortoise, and Suno AI.

Is this AI voice analyzer completely free to use online? expand_more

Yes, SmrAI offers free audio file uploads (WAV, MP3, M4A, FLAC) and real-time microphone testing without mandatory registration.

Can AI voice deepfakes bypass telephone verification systems? expand_more

Voice cloning tools can deceive humans, but SmrAI detects synthetic vocoder artifacts even through compressed 8 kHz telephony codecs.

Are uploaded voice recordings saved or shared? expand_more

No. Audio buffers are processed ephemerally in RAM and immediately deleted post-analysis under our Zero Data Retention policy.

Need to Inspect Deepfake Images or Synthetic Text?

SmrAI provides complete multi-modal authenticity verification across text, voice, and vision.

Image Vision Detector Text AI Detector