Modulate Raises $25 Million to Advance Audio-Native Deepfake Detection
Frontier AI firm Modulate has secured an additional $25 million in funding from Future Ventures, with participation from Hyperplane and Lakestar, bringing its total capital raised to $60 million. The company specializes in audio-native AI models designed to interpret the subtleties of human speech, whether genuine or synthetically generated, without relying on transcription.
At the core of Modulate’s technology is its proprietary Ensemble Listening Model (ELM) architecture, which powers the Velma platform. Velma integrates over 100 specialized AI models, each trained to detect nuanced signals such as emotion, tone, intent, synthetic speech patterns, and conversational behavior. This multi-model approach enables real-time analysis of voice data for signs of manipulation or abuse.
According to Modulate, its platform currently processes more than 10 million hours of audio each month, with a cumulative total exceeding 600 million hours analyzed to date. The company states that both its transcription and deepfake detection models have ranked #1 on public benchmarks, including those hosted on Hugging Face.
Velma Platform Delivers Superior Accuracy and Real-Time Intervention
Modulate claims its Velma platform achieves twice the accuracy of traditional large language models (LLMs) in detecting voice-based threats while generating seven times fewer false positives. Because Velma operates in real time, it allows applications not only to detect malicious activity as it occurs but also to intervene during the conversation — a critical advantage in scenarios where delays could lead to harm.
The platform is being deployed across a range of high-impact use cases: protecting healthcare institutions from deepfake social engineering attacks, enhancing the emotional intelligence and empathy of voice AI agents, reducing extremism and harassment on social platforms, monitoring voice agent performance for quality assurance, identifying and stopping child grooming attempts conducted via voice, and shielding agents in high-risk environments from voice masking techniques used to conceal identity.
"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," explains Carter Huffman, CEO and co-founder at Modulate. "We’re already using audio-native AI to protect organizations from deepfake attacks, help voice agents understand emotion and respond with more empathy, identify dangerous behavior in online conversations, and monitor whether voice agents are actually performing the way they’re supposed to."
Funding to Scale Team, Infrastructure, and Developer Access
The newly raised $25 million will be allocated toward expanding Modulate’s engineering and research teams, scaling its infrastructure, and broadening access to its models, APIs, SDKs, and integration tools for developers and partners building voice-enabled applications. The investment also supports continued advancement of the company’s underlying research in audio-native AI.
Huffman emphasized that the goal is to eliminate the need for developers to reconstruct audio intelligence layers from scratch when creating new voice experiences. "Our mission is to build the models and infrastructure that let them focus on the application they want to create," he stated. "The opportunity facing audio-native AI is expanding incredibly quickly. We’ve built the technology and proven it at scale, and this investment lets us grow the team and move faster to meet that demand."
As AI-generated deepfakes become more accessible and malicious use of voice manipulation grows, Modulate positions its technology as a critical defense layer. The company asserts that detecting deepfakes and identifying abusive voice interactions in time to prevent harm is nearly impossible without AI-driven audio analysis — a capability it aims to make widely available through scalable, developer-friendly tools.
Source: SecurityWeek