Estimated reading time: 4 minutes · Last updated:
Modulate announced a $25 million raise to expand its audio-first artificial intelligence capabilities and hire engineering and infrastructure staff. The company says its systems are already applied to fraud prevention, supervising voice AI agents, trust-and-safety monitoring and customer experience work. Carter Huffman, Modulate co-founder and CEO, framed the move as a response to voice becoming a primary interface for AI and the new problems that follow. Modulate reports its models process more than 10 million hours of audio each month and have passed 600 million hours in total, figures it cites as evidence of traction.
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.
Carter Huffman, Modulate co-founder and CEO
Key takeaways
- Funding: Modulate raised $25 million to expand engineering, infrastructure and partner support for its audio-focused AI.
- Scale: The company says its models analyze more than 10 million hours of audio per month and have processed over 600 million hours in total.
- Use cases: Modulate positions its technology for fraud prevention, voice-AI supervision, trust and safety, and customer-experience monitoring.
- Fraud context: The FBI's 2025 Internet Crime Report recorded about 22,000 AI-related fraud complaints with losses exceeding $893 million.
Table of contents
- Key takeaways
- Why Modulate is raising fresh capital now
- How Modulate describes its technology and scale
- Where this sits in the wider fraud and safety picture
- What Modulate must deliver next to justify the round
- Near-term case for and against Modulate’s expansion
- What to be careful about
- Frequently asked questions
Why Modulate is raising fresh capital now
Modulate’s $25 million round is presented as a growth-stage infusion to expand staff and cloud or on-premise capacity so partners can embed its audio models. The company told investors and customers that demand for voice-aware safeguards has risen as more firms add voice interfaces and conversational agents.
Carter Huffman, Modulate’s co-founder and CEO, framed the need as technical: voice interactions carry information—emotion, timbre, background cues—that a text transcript omits. The firm says that gap creates new security and safety challenges, and it plans to use the funding to broaden engineering and operational resources that support developers and enterprise partners building voice products.
How Modulate describes its technology and scale
Modulate says it runs more than a hundred specialised models that operate together to analyse audio with lower compute and cost than larger general-purpose models. That architecture is presented as an audio-native design: models tuned to sound features rather than relying solely on text transcriptions.
On volume, the company reports analysing over 10 million hours of audio every month and says total audio processed has passed 600 million hours. At the cited monthly run-rate, that total implies several years of cumulative processing work at scale; the figure is offered to underline both training data depth and operational footprint.
Where this sits in the wider fraud and safety picture
The fundraising arrives amid rising use of synthetic audio in criminal schemes. The FBI’s 2025 Internet Crime Report recorded roughly 22,000 AI-related fraud complaints and more than $893 million in losses, and industry research cited by the company shows large firms reporting a high rate of attacks tied to AI-created content.
Congressional researchers cited in coverage estimate that more than 95% of victims of voice cloning do not report losses, a gap that firms like Modulate argue increases demand for automated detection and monitoring. PYMNTS Intelligence reported that 58% of companies with at least $1 billion in annual revenue encountered AI-generated documents or deepfake attacks in the past year, highlighting enterprise exposure.
What Modulate must deliver next to justify the round
For the investment to pay off, Modulate will need to translate detection capabilities into reliable, low-cost services that integrate with security workflows and contact-center stacks. The company itself flags partner demand across security, communications and AI-agent supervision, and the raise is explicitly to add people and infrastructure to meet that demand.
A parallel technical challenge is the detection arms race: as audio-synthesis tools improve, detection models must evolve quickly to avoid degradation. Modulate’s claims about model breadth and historical processing volume are relevant here, but buyers will want independent validation, documented performance metrics and a clear integration roadmap before committing at scale.
Near-term case for and against Modulate’s expansion
The case for
- Growing incidence of AI voice fraud and enterprise adoption of voice agents create immediate commercial demand for audio detection and supervision.
- Reported processing scale—more than 10 million audio hours monthly and over 600 million hours total—gives Modulate data depth to refine models and reduce false positives.
The case against
- Detection must keep pace with rapidly improving synthesis tools; a lag would reduce product effectiveness and customer trust.
- Buyers may require third-party performance audits and enterprise-grade SLAs; failing to provide those quickly could slow procurement.
What to be careful about
- An accelerating arms race between synthetic-audio generation and detection could erode model accuracy unless research and retraining keep pace.
- Under-reporting of voice-cloning incidents (estimated as more than 95% unreported) obscures the true market size and may delay enterprise investment.
- Claims about cumulative hours processed and monthly ingest require independent verification before they can be treated as proof of robustness.
- If infrastructure expansion lags partner demand, integration projects and pilot programs could stall, slowing revenue growth.
The bottom line
The $25 million round positions Modulate to scale operations as demand for audio-aware safeguards grows. The company’s stated monthly ingest and cumulative-processing figures are sizable and, if verified, would support model training and continuous updates. Delivering enterprise-ready integrations, published performance metrics and rapid model retraining will determine whether the investment accelerates adoption or meets resistance from buyers seeking independent validation. For now, the raise signals investor confidence in demand for audio-native AI, but execution will be the decisive factor.
What to watch
- Watch for Modulate’s next product roadmap update or model-release announcement; no date has been set.
- Watch for independent audits or benchmark reports validating Modulate’s detection accuracy; no date has been set.
Frequently asked questions
How much did Modulate raise and what will the money be used for?
Modulate raised $25 million and said the funds will be used to expand engineering, infrastructure and partner support to meet rising demand for audio AI across security, agent oversight, trust-and-safety functions and customer experience.
How large is Modulate’s stated dataset or operational footprint?
The company reports processing more than 10 million hours of audio each month and says total audio processed has passed 600 million hours.
How does this funding relate to AI-driven fraud trends?
The raise comes as agencies and industry bodies report rising synthetic-audio abuse: the FBI’s 2025 Internet Crime Report cites about 22,000 AI-related fraud complaints and over $893 million in losses, a backdrop Modulate points to when selling detection services.
Related reading
