Voxmind, a London-based voice biometrics and deepfake detection startup, has closed a £550,000 pre-seed round led by Ascension Ventures. The round also includes participation from Mark McDermott, co-founder of ScreenCloud, lead angel Russell Hart, and members of the Cambridge angel network.
The funding arrives at a structural inflection point in enterprise authentication. Microsoft retired Azure Speaker Recognition in September 2025 and AWS ends support for Voice ID in May 2026, meaning two of the three major cloud providers have now exited voice biometrics. Hardware OEMs, contact centre operators, and enterprises that relied on those services are now without a long-term authentication layer, precisely as deepfake voice fraud accelerates against financial services, telecoms, and enterprise voice channels.
Founded in January 2024, Voxmind has developed a patent-pending phoneme-frequency extraction engine that analyses the biomechanics of the human vocal tract - physical signal characteristics governed by anatomy, not language. The result is voice authentication and deepfake detection that works across all languages by design, achieves 99.8% deepfake detection accuracy in under three seconds, and runs in under 500MB of runtime memory with no GPU or cloud connectivity required. The architecture supports three deployment models: an on-device OEM SDK for hardware manufacturers; platform-native integration for CCaaS and UCaaS operators via WebSocket, gRPC, SIP/SIPREC, and AudioHook; and a cloud API for enterprise and fintech deployments.
Voxmind has already signed an OEM agreement with a major unified communications hardware provider, embedding its voice biometric SDK directly into enterprise IP phone hardware. Additional commercial pipeline spans community banking in the United States, telecoms in Europe, and strategic partnership conversations across global IT services and contact centre operators.
The pre-seed capital will be deployed across commercial sales into financial services, telecoms, and contact centre operators; model optimisation for edge deployment; and ISO 27001 certification on an eight-month delivery timeline.
Founder and CEO Jai Keerthi brings fifteen years of experience designing mission-critical infrastructure across energy, finance, retail, and utilities, including SCADA systems, trading platforms, and high-availability environments, with deep expertise in AWS, fraud prevention, and systems where downtime is not an option. At Voxmind, he leads product strategy, enterprise partnerships, and the technical roadmap, from the Tier 1 UC OEM integration through to edge deployment and the commercial pipeline.
Voice is the last unencrypted frontier in enterprise security, and generative AI just made the stakes existential. Voxmind's physics-based approach, on-device deployment, and existing OEM footprint give them a structural advantage that's very difficult to replicate. Excited to lead this round and back the team as they scale.
Every major enterprise that relied on cloud voice APIs is now exposed precisely as AI voice fraud accelerates. We built for this architecture before the gap existed, physics-based, on-device, deployable anywhere. Now we just need to fill it.
