Resemble AI

Resemble AI is a generative AI security platform that helps enterprises create, verify, and detect synthetic voice, image, and video content. The company started in 2019 by building foundational generative voice models, and that deep understanding of how synthetic speech is produced became the basis for its deepfake detection technology, which shipped the same year. Today the platform spans three pillars: Generate (text-to-speech, voice cloning, and speech-to-speech tools), Verify (biometric identity verification and imperceptible audio, image, and video watermarking), and Detect (real-time multimodal deepfake detection tested against 160+ generative AI models). The platform's flagship detection model, DETECT-3B Omni, is a 3-billion-parameter architecture that identifies AI-generated audio, image, and video in a single unified model, returning verdicts in under 300ms with 98.1% accuracy on independent benchmarks. Resemble Meetings extends this into live calls on Zoom, Teams, Google Meet, and Webex, flagging synthetic voices in real time with a forensic audit trail. On the generation side, Resemble's Chatterbox model family (open-sourced under MIT license) powers sub-200ms text-to-speech and voice cloning from as little as 10 seconds of audio, with emotion and tone control across 100+ languages. Its PerTh watermarking technology embeds a permanent, inaudible signal into generated audio that survives compression and editing, aligning with EU AI Act Article 50 content-provenance requirements. Resemble AI is built for enterprise and regulated environments: it supports SOC 2 Type II controls, GDPR and HIPAA-aligned data handling, SSO/SAML, and fully air-gapped on-premises deployment for organizations that cannot send data off-site. Customers cited include Netflix, Paramount, Deutsche Telekom, Telnyx, and the World Bank, spanning use cases like executive impersonation defense, call-center fraud prevention, identity verification and KYC, media watermarking, and voice agents. The tool suits security, trust and safety, and compliance teams that need to detect and respond to synthetic media fraud, as well as developers and studios building voice-driven products who need production-grade cloning and TTS with built-in provenance. It is less suited to casual creators looking for a simple, self-serve consumer voiceover app, since the platform's depth and per-second pricing are oriented toward technical, API-driven, enterprise-scale deployments.