Creates a unique vocal signature from any audio sample and searches incoming audio against enrolled profiles in real time. Pairs with Resemble Detect to add synthetic verification to every identity check.
Minimum audio required to enroll a speaker profile.
RESEMBLE IDENTITY ATTACK COVERAGE
Can a voice clone fool the system? We verify the signature, not just the sound.
Voice impersonation attempt
FLAGGED
FLAGGED
Known bad actor using a different name or account
FLAGGED
FLAGGED
Replay attack (pre-recorded authorized speaker)
FLAGGED
FLAGGED
Legitimate caller, known enrolled profile
VERIFIED
VERIFIED
AI voice clone of an enrolled speaker
FLAGGED
FLAGGED
OUR CAPABILITIES
Verify identity continuously, without adding steps for callers or agents.
Built for real world audio. Handles telephony, compression, and noisy environments natively. No preprocessing or clean audio required.
Enroll and verify
Create a speaker profile from 4 seconds of audio. It returns a match distance score on every subsequent call, and adapts to voice changes over time without re-enrollment.
Add to fraud watchlist
Enroll known bad actors once and get alerted whenever that voice appears on any channel, regardless of name or account.
Pair with Resemble Detect
Identity confirms who is speaking. Detect confirms the audio is real. Together they form the voice layer in a face and voice fraud prevention stack, replacing multiple vendors with one platform and one API.
1,000,000+
Speaker profiles supported. Enroll once.
$0.0005
Per identity search, pay-as-you-go
99%
Uptime SLA on enterprise plans
AI identity RISK BY INDUSTRY
An unverified voice is an unlocked account.
The attack is the same. The consequences depend on where it happens.
Banking & Finance
Industry risk:
Voice clone fraud bypassing contact center authentication before transactions are approved.
Resemble AI solution:
Verify caller identity in real time. Flag impersonators before access is granted.
Legal & Judicial
Industry risk:
Audio evidence provenance unverifiable without chain-of-custody documentation.
Resemble AI solution:
Confirm speaker identity in recordings and support chain-of-custody requirements.
Telecommunications
Industry risk:
Vishing attacks exploiting agent trust on inbound calls with no real-time signal.
Resemble AI solution:
Alert agents instantly when a call shows signs of voice fraud or a watchlisted speaker.
Media & Entertainment
Industry risk:
Unauthorized cloning of protected talent voices in production or distribution.
Resemble AI solution:
Enroll talent profiles and detect unauthorized voice use without manual review.
Healthcare and HealthTech
Industry risk:
Unverified synthetic instructions in clinical pathways.
Resemble AI solution:
Verify clinical voice instructions are authentic before they reach care teams.
Corporate Security
Industry risk:
Known social engineers re-contacting the organization under different identities.
Resemble AI solution:
Enroll flagged speakers once. Catch repeat contact attempts across every channel.
4 seconds of audio at a minimum sample rate of 16 kHz. Profiles capture pitch, tone, resonance, and cadence from a short sample submitted via audio URL.
Does more audio improve enrollment quality?
Yes. Concatenating multiple clips from different channel conditions into a single file creates a more robust speaker profile. The system automatically segments and enrolls a combined file, so there is no need to submit clips individually.
What does the identity search return?
A match distance score per enrolled speaker, identifying the closest match or confirming no match was found. It is not a binary yes or no, the score gives you control over your own threshold.
Can it detect AI voice clones trying to impersonate an enrolled speaker?
Yes, but for the most robust protection, pair Identity with Resemble Detect. Identity confirms whether the voice matches a known profile. Detect confirms whether the audio is real or synthetically generated. Both checks run through the same API.
How does the fraud watchlist work?
Enroll known bad actors as flagged profiles. Every inbound call and audio submission is searched against the watchlist. An alert fires when a match is found, regardless of the caller ID, name, or account used.
Does audio leave our infrastructure during processing?Does audio leave our infrastructure during processing?
Not with on-premises deployment. Voice profiles are encrypted and stored on your own infrastructure. Cloud API processing is available for organizations without strict data residency requirements.
What are the enterprise deployment options?
Cloud API, on-premises, and air-gapped environments. On-prem deployments are shipped as containerized images with updates included throughout the contract. SLA documentation available on request.
Get complete generative AI security
Join thousands of developers and enterprises securing with Resemble AI
