Modulate Raises $25M for Boston Voice Deepfake AI

Boston’s Modulate raised $25M led by Future Ventures to scale audio-native AI for deepfake detection and voice-agent monitoring. Funding totals and Velma.

Updated

Share this article

Home › Boston News

Boston voice intelligence startup Modulate has raised $25 million to scale small-model voice AI for deepfake detection and AI-agent compliance, AI Insider reported on Sept. 30, 2026. Future Ventures led the round, with participation from Hyperplane and Lakestar, according to that account.

Part of our Boston News coverage. Browse more in Housing, Development and Business.

Advertisement

Founded in 2017 by Mike Pappas and Carter Huffman—who met as MIT physics undergraduates—Modulate began in gaming voice modulation, moved into voice moderation, and now focuses on spotting synthetic audio and analyzing spoken intent for enterprises such as call centers.

Modulate $25M Round: Product and Hiring Plans

AI Insider reported that Modulate’s platform runs more than 100 small models in two groups: signal-extraction models that assess emotion, tone, language and synthetic voices, and analysis models that flag intent, rule violations and possible scams. Huffman said the small-model approach avoids specialized hardware and heavy compute while making it easier to train new detectors an orchestrator can call on demand.

The company employs about 40 to 45 people and plans to hire roughly 10 more for model building. It is also expanding on-premises and on-device deployments for privacy-sensitive customers. PitchBook data cited in the report showed prior fundraising of about $41 million at a $170 million valuation.

Why Voice Deepfakes Matter to Boston Enterprises

Call centers and regulated industries in Greater Boston face rising risk from cloned voices and unsupervised AI agents on customer lines. Modulate’s pitch—detect synthetic speech and monitor agent quality and compliance—targets that operational gap. Huffman also cautioned that polite tone toward an AI agent can mask deep customer dissatisfaction, a nuance enterprises often misread when they lean only on sentiment scores.

The raise adds to a busy autumn for Boston AI companies. Compare Modulate’s voice focus with our coverage of the Harvey Boston office legal-AI expansion and the local launch of Legato AI hearing glasses.

Our Take

Modulate’s small-model stack is a bet that specialized, orchestrated detectors will beat one giant model for regulated voice workflows. If the hiring plan sticks, Boston gains another mid-stage AI employer competing for speech and safety talent against Cambridge labs and larger platforms. The practical test will be whether on-prem deployments win banks and health systems that will not send call audio to the public cloud.

Primary sources: AI Insider on Modulate’s $25M raise; company background via founder statements in that report; Modulate (company site).

Advertisement

Funding totals and who is running the company

Modulate’s own announcement, dated Sept. 28, says the round brings its total funding to $60 million. That differs from the roughly $41 million prior figure PitchBook data showed in AI Insider’s account, so treat the totals as reported by different sources. SiliconANGLE adds that Hyperplane backed Modulate’s $2 million seed round and Lakestar led its $30 million Series A in 2022. The Next Web reports that Carter Huffman is now chief executive and Mike Pappas is chairman.

What Velma does

Modulate’s main product, Velma, analyzes audio directly rather than working from transcripts. According to The Next Web, it picks up emotion, tone, intent, emphasis and whether a voice is synthetic, then combines those signals to flag higher-level events such as a fraud attempt, harassment or a frustrated customer, in real time during a call. Modulate says its underlying Ensemble Listening Model uses more than 100 specialized models, and claims Velma has “2x greater accuracy than traditional LLMs” at detecting true positives and 7x fewer false positives; SiliconANGLE reports the company says the design is up to 1,000 times more efficient than handing audio to one large model. These are company claims.

Scale and the use of the new money

In a post from the co-founders, Modulate says its models analyze more than 10 million hours of audio each month and have processed more than 600 million hours in total. Huffman said, “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.” The money will go toward research, engineering, new SDKs and APIs, and partner integrations so developers do not have to rebuild an audio intelligence layer for every voice product. Future Ventures co-founder Steve Jurvetson said Modulate “has gained a significant technical lead in audio-native AI.”

Sources

Further Reading

Read next in Artificial Intelligence

Get The Boston Report Briefing

Our free weekly email with the most important stories from Boston and the world. Unsubscribe anytime.

Free. We never sell your address. Privacy policy

Get The Boston Report Briefing
Our free weekly email with the most important stories from Boston and the world. Unsubscribe anytime.
Sign up