Boston-based voice intelligence startup Modulate has secured $25 million in new funding for its platform, which combines many small models to provide transcription, emotional analysis, deepfake and AI-music detection, and policy enforcement for voice agents in regulated industries.
Funding and investors
The round was led by Future Ventures, with participation from Hyperplane and Lakestar. According to PitchBook data, Modulate had previously raised $41 million and was valued at $170 million before this latest round.
Origins and strategic shift
Modulate was founded in 2017 by Mike Pappas and Carter Huffman, who met as undergraduates studying physics at the Massachusetts Institute of Technology (MIT). The company began by offering voice modulation for gaming but later moved into voice-based moderation tools. With the rise of voice AI models, Modulate has shifted focus toward detecting AI-generated audio and analyzing the intent behind what people say.
Technology and model approach
The company runs more than 100 models, broadly split into two categories:
- Signal extraction models: designed to capture vocal features such as emotion, tone, language, and whether a voice is synthetic.
- Analysis/detection models: intended to infer intent, determine if callers are breaking rules, or identify scam attempts.
Carter Huffman explained that running many smaller models means Modulate does not rely on specialized hardware or massive compute, which matters if token costs rise. Smaller models are also easier to train, add to the suite, and invoke through an orchestrator when needed.
Customers and use cases
Modulate serves a varied customer base but specializes in deepfake detection and alerting organizations like call centers about potential scams. The company also monitors how AI agents interact with customers to assess call quality and to ensure compliance in regulated areas. In many deployments, Modulate sits alongside the company’s voice stack specifically to analyze calls.
As enterprises increasingly adopt AI-powered customer service, understanding why a call succeeds or fails becomes important beyond basic transcription. Huffman noted that simple emotion labels can be misleading — a neutral or polite-sounding customer can still be dissatisfied — and Modulate aims to provide granular insights into intent and customer response.
Security monitoring and product direction
Modulate’s technology is also used to detect voice-based cyberattacks. The startup is working to expand its ability to deploy on-premises and on devices to meet privacy requirements.
Team and next steps
The company currently employs 40–45 people and plans to hire about 10 more in the coming months to strengthen model development. The new funding will support further development of detection, intent analysis, and on-prem/on-device deployment capabilities as Modulate competes in a growing field focused on human-like AI voices and voice-based security.
