Modulate Raises $25M to Scale Deepfake Voice Detection

Modulate announced a $25 million investment led by Future Ventures, with participation from Hyperplane and Lakestar, taking the company’s total funding to $60 million. The company said the capital will support growth of its team and infrastructure and expand tools for developers building voice applications.

The funding is intended to broaden the models, application programming interfaces, software development kits, integrations and deployment options that sit on top of Modulate’s Velma platform. Modulate describes Velma as an audio-native system that analyzes voice signals rather than generating speech.

Velma is built on a proprietary Ensemble Listening Model architecture that combines many narrow models each tuned to specific audio signals. The platform includes more than 100 specialized models that detect features such as emotion, tone, intent, synthetic speech and conversational behavior. Modulate reports Velma processes over 10 million hours of audio per month and has analyzed more than 600 million hours in total.

The company reports its transcription and deepfake detection models ranked first on public benchmarks. Modulate also reports that Velma yields higher measured accuracy and fewer false positives than standard large language models when those models are applied to audio tasks, based on internal evaluations.

Because Velma operates in real time, Modulate says applications can flag or intervene during live conversations. The company lists use cases including protecting healthcare institutions from voice impersonation attacks, improving emotion and empathy in voice agents, reducing harassment and extremist content on social platforms, monitoring voice agent performance, detecting child grooming, and masking agent identity in high-risk scenarios.

Carter Huffman, CEO and co-founder, described the challenge: “Voice is becoming a primary interface for AI, and that creates a whole new set of problems that cannot be solved from a transcript.” He added the company aims to provide the audio intelligence layer so developers can focus on their applications.

Modulate positions Velma as a detection and monitoring layer that can be integrated into voice-based systems to flag suspicious audio in real time amid rising concern over malicious synthetic voice and other audio deepfakes.