Modulate Secures $25M for AI Fraud Prevention
Modulate raised $25 million in funding to scale its audio-native AI technology, which detects deepfakes and fraud in voice systems, as businesses face
Modulate secured $25 million in new funding on September 28 to advance its audio-native AI capabilities for fraud prevention and deepfake detection. The round was led by Future Ventures, with participation from Hyperplane and Lakestar, bringing the company's total funding to $60 million. The capital will fuel AI research and product development.
Carter Huffman, Modulate's co-founder and CEO, stated the growing need for such tools. "Voice is becoming a primary interface for AI, and that creates a whole new set of problems that can’t be solved from a transcript," he said. The company reports growing adoption of its technology across fraud prevention and online trust and safety.
Technology capabilities and market position
Modulate's audio analysis platform uses specialized models to detect deepfakes and fraud with superior accuracy and real-time performance. Its models have processed more than 600 million hours of audio in total and now analyze over 10 million hours every month. The company's deepfake speech detection technology holds the top position on a Hugging Face public benchmark.
The core platform, called Velma, is powered by an Ensemble Listening Model (ELM) architecture that coordinates more than 100 specialized audio models. According to Modulate, this approach delivers twice the accuracy of traditional large language models when detecting true positives and generates seven times fewer false positives. The company claims the technology achieves 98.9% accuracy on public benchmark data and operates in real time, allowing for intervention during ongoing conversations. Modulate said ELM has demonstrated as much as 1,000 times greater efficiency than a single large-model approach, reducing the computing power, memory, and cost required.
Application areas and use cases
The technology is deployed across healthcare fraud prevention, voice agent supervision, and trust and safety applications for businesses. Modulate's models are used to guard healthcare institutions from deepfake attacks, recognize emotion in AI agents to enhance empathy, and monitor voice agent performance. They also identify harmful behavior such as harassment, child grooming, and extremism on social platforms.
The broader goal is to create an audio intelligence layer that developers can integrate into voice agents, security products, and communications platforms. This allows businesses to access specialized audio analysis without building the models themselves. Demand comes from partners building applications across security, customer experience, communications, and AI agent supervision.
Developer accessibility and expansion
Modulate plans to use the funding to make its capabilities more accessible through expanded APIs, SDKs, and developer partnerships. The company will expand APIs, models, and deployment options for developers building voice applications. It is also growing its team and infrastructure to meet rising demand.
The move aims to support broader integration of audio intelligence. Velma combines audio signals to identify higher-level events like suspected fraud, AI agent failures, and policy violations. The company's transcription API, which recently ranked first on Hugging Face’s Open ASR Leaderboard, is priced at 3 cents per hour for batch processing. Modulate's expansion of developer tools and APIs is designed to support broader voice application integration.
