Modulate Raises $25 Million to Build Voice AI Audio Intelligence Layer

What changed

Modulate raised $25 million in funding led by Future Ventures to build an audio intelligence layer for voice AI. Its Velma system orchestrates more than 100 voice-specialized models on raw audio, with APIs aimed at detecting fraud, deepfakes, agent failures, churn signals and abuse.

Why it matters

The bet is that voice systems need to understand more than words. Transcripts can capture what was said, but they may miss signals tied to frustration, fraud or abusive behavior. Modulate is positioning raw audio analysis as a control layer for teams deploying voice agents and customer-facing systems.

The new funding is intended to expand Modulate’s models, deployment options and APIs. That could make it easier for builders to add monitoring without assembling a separate stack for every risk.

The practical test will be whether those tools work reliably outside demos. The reporting provides no funding terms, valuation, customer adoption, accuracy figures or deployment timeline, so the size of the round shows ambition more clearly than traction.