OmniSpeech
University of Maryland spinout whose zero-shot model flags AI-generated voice deepfakes in real time inside Zoom calls.
Visit Website ↗ + Add to Compare Claim This CompanyOverview
OmniSpeech is a speech-technology company whose AI Deepfake Detection product identifies AI-generated and manipulated speech in near real time, shipping as a Zoom app and Chrome extension so it can flag synthetic voices directly inside live meetings and calls. The company also sells AI noise-suppression and speech-biomarker technology built on the same underlying acoustic-phonetics research base.
The detection model is described as zero-shot and data-independent: rather than matching audio against a library of known synthetic-voice examples, it is built to flag characteristics of synthetic speech it has not specifically been trained on, which the company positions as more resilient to new voice-cloning techniques.
Founded by speech scientist Dr. Carol Espy-Wilson and based in College Park, Maryland, OmniSpeech is a University of Maryland VentureAccelerator graduate and debuted its real-time detector at CES 2026. It received a $500,000 Pre-Seed Builder Fund investment from Maryland’s TEDCO in late 2025 and is seeking additional seed funding.
Innovation Matrix Assessment
A long-running University of Maryland speech-technology venture that moved quickly from research to a shipping Zoom Marketplace deepfake detector and a CES 2026 launch.
Embedding detection directly inside the meeting platform where voice-phishing and deepfake calls actually occur is operationally more useful than a standalone scanning tool.
Only a $500K pre-seed investment disclosed so far, and the company is still actively fundraising; no named enterprise customers were found, only research and technology partners (NSF, Intel, NVIDIA).
A zero-shot, data-independent detection approach is a real technical choice that differs from library-matching deepfake detectors, though impact is unproven at scale.
University research pedigree and named technical partners lend some credibility, but no independent third-party test results were found; vendor claims of outperforming competitors on error rates are unverified.
Real-time voice-deepfake detection addresses a fast-growing social-engineering vector (voice phishing, executive impersonation) that is likely to keep mattering.
Why CISOs Should Care
Flags AI-cloned voices during live calls and meetings, directly addressing voice-based social-engineering and executive-impersonation fraud.
What Makes It Different
Uses a zero-shot, data-independent detection model rather than matching against a library of known synthetic-voice samples.
The Matrix Verdict
48/100 — EMERGING / UNRANKED
An Emerging/Unranked company: technically credible, university-rooted deepfake detection, but still at a pre-seed stage without independent efficacy validation or disclosed customers.
Editorial Note: Claims vs. Verified Findings
Performance superiority claims (MOS scores, error rates) are vendor-published; the pre-seed funding amount and founding details are corroborated by TEDCO and University of Maryland sources.
Sources
Alternatives to OmniSpeech
Adaptive Security
AI-driven platform that simulates deepfake, voice, and multichannel social-engineering attacks to train and test organizations against next-generation phishing.
Quilr
Early-stage agentic AI security startup building a 'Service-as-Software' platform to guard against human-related breaches and secure AI agent…
Zenity
Governance and security platform for AI agents and low-code/no-code development, securing agent identity, permissions and behavior across the…
Gray Swan AI
Carnegie Mellon-linked startup building automated red-teaming and runtime guardrails that frontier AI labs use to stress-test their models…
Tenzai
An agentic AI penetration testing startup building autonomous 'AI hackers' to find and validate exploitable vulnerabilities at a…
Reco
Reco secures the "agentic ecosystem" — mapping what AI agents can access across SaaS and enterprise apps, detecting…